{ "measurement": "What a trending 817-skill pack costs at startup, and why this site's own published skills curve under-predicts it by nearly five times", "ran_on": "2026-08-20", "claude_code_version": "2.1.236", "claude_code_version_recorded_at_runtime": "2.1.236 (Claude Code)", "pack": "mukul975/Anthropic-Cybersecurity-Skills", "pack_commit": "4c0b700ac5d280ba46695062077f0fe922ce3602", "not_an_anthropic_repository": "Despite the name, this is not published by Anthropic. The marketplace owner is an individual account. Stated because a reader searching for an official pack will find this one first.", "prompt": "Reply with only: OK", "population_and_estimator": { "population": "The pack exactly as it ships, copied into an otherwise empty project's .claude/skills/, measured as whole-session context against a floor re-measured in the same round. One machine, one release.", "estimator": "Mode of the per-round deltas, computed by the shared guarded estimator that refuses a no-repeat or tied mode. ONE estimator and ONE population for every cell below." }, "headline": "A trending 817-skill pack costs about 56,600 tokens on every session, 69.3 per skill, from 24 readings that agree within 48 tokens in 21 of them. This site's own published curve predicts 11,711 for 1,000 synthetic skills and re-measures at 11,701, so the published cell holds; it is simply a bad predictor of a real pack, by 5.9x per skill. WHY is NOT established. Description length, description string-uniqueness and extra frontmatter were each tested and refuted, and the remaining candidate, the description text itself, is not supported by these arms: the synthetic descriptions are provably not in the measured context at this skill count, so the synthetic arms price nothing about description text at all.", "results": { "the_pack": { "skills": 817, "description_chars": { "min": 97, "median": 396, "mean": 383, "max": 1016, "total": 312905 }, "over_documented_1024_limit": 0, "over_measured_1400_context_cut": 0 }, "measured_three_times_by_three_harnesses": { "harness_1_dedicated": { "rounds": 10, "mode": 56622, "estimator_refused": null, "min": 56622, "max": 56622, "spread": 0, "all_rounds": [ 56622, 56622, 56622, 56622, 56622, 56622, 56622, 56622, 56622, 56622 ], "per_skill": 69.3 }, "harness_3_interleaved_with_stripped": { "rounds": 6, "mode": 56592, "estimator_refused": null, "min": 17352, "max": 57489, "spread": 40137, "all_rounds": [ 17352, 56592, 56592, 56592, 17352, 57489 ], "per_skill": 69.27 }, "harness_4_interleaved_with_real_desc": { "rounds": 8, "mode": 56602, "estimator_refused": null, "min": 56602, "max": 56640, "spread": 38, "all_rounds": [ 56602, 56602, 56602, 56602, 56602, 56602, 56602, 56640 ], "per_skill": 69.28 }, "agreement": "24 readings, three modes: 56,622, 56,592 and 56,602. NOT three independent instruments, and an earlier version of this corpus said they were. All three harnesses `import skillpack_manifest_run as H` and take their reading from the same H.call(): same argv, same prompt, same floor procedure, same token expression. They were written within nine minutes of each other on the same day. Three wrappers around one measurement function agreeing is one instrument run three times, which is a repeatability check and not corroboration.", "the_actual_bands": "21 of the 24 readings sit within 48 tokens (56,592 to 56,640). 22 sit within 897 because of fmcharge round 5, and that round is the reverse of what an earlier version of this corpus said. Read as ABSOLUTE context the arms did not move at all in round 5: the full arm read 79,373 exactly as in rounds 1, 2 and 3, and the stripped arm read 79,381 exactly as in all six rounds. Only the floor moved, 22,781 down to 21,884. So the pairing INJECTED 897 tokens of floor noise into both deltas rather than removing any. That corroborates what this site already published from the plugin corpus, that pairing against an independently noisy floor adds variance here. An earlier version of this corpus claimed 22 within 48 tokens, which reached 48 only by silently dropping that third reading.", "the_one_anomaly": "2 of the 24 readings came in at 17,352 instead of about 56,600. Both are in harness 3 and both in its `full` arm, which is one half of the frontmatter comparison below: 2 of 6 there against 0 of 6 in `stripped`. So the anomaly is arm-correlated and it lands inside, and breaks, that comparison in those two rounds. It is NOT floor drift: in those rounds the floor read 22,781, identical to the rounds either side, so it is the full arm's absolute context that collapsed. Arm order within a round is fixed, full always first, and both anomalies landed in that first slot; 2 of 6 against 0 of 6 does not establish a position effect and none is claimed. The cause is NOT established. Every round of every harness is published in the `runs` array below, so the readings can be inspected rather than taken on trust. It is reported because a 3.3x low reading on a byte-identical fixture is exactly the kind of thing a corpus should not smooth away." }, "what_explains_it": { "note": "Each arm below changes ONE thing against the arm above it. Every arm is 817 skills measured against a same-round floor.", "synthetic_identical_descriptions": { "rounds": 6, "mode": 9952, "estimator_refused": null, "min": 9952, "max": 9952, "spread": 0, "all_rounds": [ 9952, 9952, 9952, 9952, 9952, 9952 ], "per_skill": 12.18, "what_it_is": "817 skills, 396-character descriptions, every description the SAME string. This reproduces the method behind this site's own published count_curve, whose fixture returns one repeated prose block." }, "synthetic_unique_but_small_vocabulary": { "rounds": 6, "mode": null, "estimator_refused": "no value repeated across 6 rounds, so the mode would be an arbitrary pick", "min": 11047, "max": 11108, "spread": 61, "all_rounds": [ 11054, 11070, 11074, 11088, 11047, 11108 ], "per_skill": null, "what_it_is": "Same, but every description a distinct string drawn from a 30-word vocabulary. Tests whether STRING uniqueness is the term.", "verdict": "Uniqueness alone moves it about 1,100 tokens. Refuted as the explanation. But see synthetic_descriptions_are_not_in_context below before reading either synthetic arm as a measurement of what descriptions cost." }, "real_descriptions_everything_else_synthetic": { "rounds": 8, "mode": 44572, "estimator_refused": null, "min": 44572, "max": 44572, "spread": 0, "all_rounds": [ 44572, 44572, 44572, 44572, 44572, 44572, 44572, 44572 ], "per_skill": 54.56, "what_it_is": "817 skills, 200-byte bodies, name and description only, no reference files, no scripts, no extra frontmatter keys, carrying the pack's REAL descriptions. 6,156 distinct word types in the real set.", "fixture_control_failed_deliberately_reported": "This arm's own fixture control marked every round `failed` and the rounds are published with that status intact. The CAUSE was initially misdiagnosed here as YAML newline normalisation losing 3.9% of the characters. It is worse than that. Rewriting each description as an unquoted single-line scalar produces INVALID YAML for exactly 22 of the 817, because those descriptions contain a colon-space (for example 'CMMC Level 2 certification: scope CUI and FCI'). Those 22 files lose BOTH name and description and would not register as skills at all. They are also the long ones, mean 553 characters against the pack's 383, and they account for all 12,163 of the missing characters exactly. So this arm is 795 usable skills plus 22 broken files, not 817 skills at 96.1% of the text, and its per-skill divisor in the run rows is wrong. No share of the total is claimed from it." }, "real_files_with_extra_frontmatter_stripped": { "rounds": 6, "mode": 56600, "estimator_refused": null, "min": 56600, "max": 57497, "spread": 897, "all_rounds": [ 56600, 56600, 56600, 56600, 56600, 57497 ], "per_skill": 69.28, "what_it_is": "The real files, byte-identical bodies and reference files, with every frontmatter key EXCEPT name and description removed: 338,442 bytes of tags, MITRE/NIST mappings, author and license deleted.", "verdict": "Read as ABSOLUTE context the two arms are perfectly reproducible: the full arm returned 79,373 in every non-anomalous round and the stripped arm 79,381 in ALL six, each with a spread of exactly zero. So stripped minus full is +8 tokens, and that +8 is DETERMINISTIC rather than noise: it reproduces exactly, four rounds out of four usable. Deleting 338,442 bytes of frontmatter made the reading 8 tokens MORE expensive, a systematic offset far too small to be a cost for that much text. The extra frontmatter is not charged, and Anthropic's documentation, which says Level 1 loads `name` and `description`, is accurate on these rounds. Two caveats: the null rests on 4 of 6 usable rounds because the anomaly landed on the full arm twice, and the interleaving control did NOT protect this comparison when it did." } }, "what_this_site_predicted": { "count_curve_1000_skills": 11711, "desc_at_scale_600_skills_300_char": 9684, "count_curve_1000_re_measured": 11701, "this_sites_own_REAL_pack_corpus_already_agreed": "Important corrective to the framing. skillpack-cost-2-1-233.json measured six REAL packs and got per-skill figures of 34.08, 81.26, 95.92, 117.0, 262.0 and 274.0. The 69.30 measured here sits comfortably inside that range, and google/skills at 81.26 per skill on 111 real skills at a 397-character description median is the closest comparable cell there is. This site's REAL-pack measurements never disagreed with this result; only the SYNTHETIC count_curve does. The gap is synthetic-versus-real, not old-versus-new, and any framing that says this site was blindsided by its own data would be wrong.", "the_closest_real_comparison": "google/skills is an almost exact control for this pack: 111 skills at a 397-character description median measuring 81.26 tokens per skill, against 817 skills at a 396-character median measuring 69.30. Near-identical description length, 7.4x apart in skill count, 17% apart per skill. So REAL packs barely collapse with count. The SYNTHETIC count_curve collapses hard over the same range, from about 39 tokens per skill in its flat regime to 13.46 interpolated at 817. Two points are not a curve and none is drawn through them, but they are enough to say which published thing is out of line.", "verdict": "This site's published curve predicts 11,711 tokens for 1,000 synthetic skills at 90-character descriptions, and curve-stability-2-1-234.json re-measured that same cell at mode 11,701 over 8 rounds (spread 897), so the published cell holds. A real pack of 817 skills costs about 56,600. Per skill that is 69.30 against 11.71, a factor of 5.9; against the published curve linearly interpolated to 817 skills it is 5.15 (56,622 / 10,994). An earlier version of this corpus called it 4.8x per skill, which was the ratio of TOTALS across different skill counts and understated the site's own error. The 9,952 synthetic cell measured here is a NEW cell at 817 skills and 396-character descriptions; it reproduces nothing previously published and must not be read as a reproduction check." } }, "controls": { "paired": "Floor re-measured every round in every harness; every delta is against that round.", "interleaved": "Arms that are compared ran inside the same round, so drift cannot land on one arm.", "arrival": { "what": "Every reply must be exactly OK, published per run as reply_exact_ok. On 817 third-party descriptions entering context this doubles as a prompt-injection check.", "harness_1": "10/10", "harness_2_identical": "6/6", "harness_2_unique": "6/6", "harness_3_full": "6/6", "harness_3_stripped": "6/6", "harness_4_full": "8/8", "harness_4_real_desc": "8/8" }, "isolation": "--setting-sources project asserted onto argv in every call.", "fixture": "Skills counted back off disk every run. The stripped arm additionally hashes both arms' name and description sets and publishes names_match/descs_match, so 'only the frontmatter differs' is measured. The real-description arm hashes its set against the pack's and FAILED that check, which is why it is reported as a bound.", "no_invocation": "Trivial prompt; no skill body is ever executed or read.", "security_scanned": "817 third-party files from an individual author. No plugin hooks, no lifecycle or postinstall scripts. 1,093 shipped Python scripts, all unique rather than templated. Every danger-grep hit was read in context and was a false positive: eval( matches are YARA rules and Splunk SPL functions, curl|sh matches are grype/syft/uv install lines and detection rules that FLAG that pattern, base64 is JWT and PowerShell deobfuscation, /etc/shadow and keychain are forensic targets, and every AWS key is AWS's own AKIAIOSFODNN7EXAMPLE or a honeytoken decoy. The grep carried a positive control that fired 2/2." }, "known_limits": { "the_anomaly_is_unexplained": "2 of 24 readings at 17,352. Not explained, not excluded from the published rounds, and not used to widen the headline, which rests on the 21 readings that agree within 48 tokens. Anyone re-running this should expect to see it.", "THE_CAUSE_IS_NOT_ESTABLISHED": "A real 817-skill pack costs about 5x what this site's curve predicts, and this corpus does NOT explain why. Three candidates were tested and refuted (description length, description string-uniqueness, extra frontmatter), and the fourth, that the description TEXT is the term, is NOT supported by the arms here. An earlier version of this corpus and an unpublished draft claimed a vocabulary mechanism. It was withdrawn on the arithmetic below before publication.", "synthetic_descriptions_are_not_in_context": "The decisive check, and the reason the vocabulary story was dropped. The synthetic arms carry 817 x 396 = 323,532 characters of description, roughly 43,800 whitespace-delimited words. Any BPE tokeniser charges at least one token per word, so if those descriptions were in the measured context they could not cost less than about 43,800 tokens. The entire arm delta is 9,952 to 11,108. The synthetic descriptions are therefore provably NOT in the measured context at 817 skills, which means the identical-versus-unique comparison prices nothing about description text and its 11% gap is not an effect size. This site's own skills-desc-at-scale-2-1-231.json already says so in the other direction: at 600 skills the 300-character arm (9,684) is 778 tokens CHEAPER than the 30-character arm (10,462). Description length stops being charged well below 817 skills.", "no_arm_isolates_vocabulary": "The real-description arm changes vocabulary AND syntax AND length distribution (constant 396 characters becomes 97 to 1,016, with 22 zeros) AND provenance, all in one step. The one control that would test vocabulary alone, synthetic descriptions drawn from a ~6,000-word vocabulary at constant length, was not run. It is the obvious next measurement.", "one_pack_one_machine_one_release": "Opus on Claude Code 2.1.236, macOS. One pack. Whether every real pack behaves this way is not established; what IS established is that at least one does and that the synthetic fixture does not.", "carried_forward_one_skill_is_inside_the_noise": "skills-cost-2-1-233.json and curve-stability-2-1-234.json carry that a single skill, and up to 10, sit inside this harness family's round-to-round noise. NOT retired and NOT tested here: every cell in this corpus is 817 skills, far above that floor, and nothing here bears on small counts.", "carried_forward_bodies_are_free_until_invoked": "skills-cost-2-1-233.json measured body size at 40 skills as free (200, 4,000 and 28,000 bytes all landed together). This corpus does NOT re-verify that and does not contradict it: the arm carrying real 9,949-byte median bodies and the arm carrying 200-byte bodies differ in their descriptions too, so body cost is not isolated here. Carried forward unretired.", "carried_forward_descriptions_above_500_chars_are_cheap": "desc-limit-2-1-235.json measured 0.015 tokens per character above 500 characters at 40 skills. This corpus does not bear on it: every cell here is 817 skills, and the arithmetic above shows synthetic descriptions are not charged at all at that count. Carried forward unretired and untested.", "the_readings_are_stable_in_ABSOLUTE_context": "The cleanest way to read this corpus. Within each harness the absolute session context is byte-identical round to round: 79,401 ten times in harness 1, 79,373 in every non-anomalous fmcharge full round, 79,381 in all six fmcharge stripped rounds. The variation between the three published modes (56,622 / 56,592 / 56,602) and the one 57,489 outlier is FLOOR movement, not pack movement. The pack adds a fixed amount; the empty-project baseline is what wanders.", "runs_versus_rounds": "The `runs` array holds 80 RUNS across 30 ROUNDS, not 80 rounds: cybersec817 10 rounds, descuniq 6, fmcharge 6, realdesc 8, each round contributing one floor run plus one or two arm runs. An earlier version of this corpus and of the calculator note said 80 rounds.", "low_skill_count_rates_are_NOT_all_higher": "Recorded because a draft asserted the opposite. This site's real-pack rates at low counts do NOT all sit above the 69.30 measured here: skillpack-cost-2-1-233.json has mattpocock/skills at 34.08 per skill on 25 shipped skills, less than half of it, alongside addyosmani at 95.92 on 24 and three single-skill packs at 117.0, 262.0 and 274.0. The six measured real packs span 34.08 to 274.0 and do not order by count, so no count-times-rate arithmetic is offered for a subset of this pack.", "two_points_are_not_a_trend": "google/skills at 81.26 on 111 skills and this pack at 69.30 on 817 share a description median (397 and 396) and sit 14.7% apart, taking the larger as the base. That is TWO packs chosen because their description medians match, out of six measured, and this site's own skillpack corpus separately records that per-skill cost does NOT order by description length. So the pair is a useful coincidence and NOT evidence that real packs are flat in count. Across all six the range is 34.08 to 274.0.", "the_1400_character_cut_does_not_bite_this_pack": "desc-truncation-2-1-235.json found a description stops reaching context between 1,400 and 1,550 characters. Zero of these 817 descriptions exceed 1,024 and the longest is 1,016, so nothing here is truncated. Worth stating because Anthropic's own repository breaks the documented 1,024 limit in two of its five plugins." }, "arms": 5, "runs": [ { "harness": "cybersec817", "arm": "floor", "round": 0, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 0, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 1, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 1, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 2, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 2, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 3, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 3, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 4, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 4, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 5, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 5, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 6, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 6, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 7, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 7, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 8, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 8, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "floor", "round": 9, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": 0, "names_match": null, "descs_match": null }, { "harness": "cybersec817", "arm": "anthropic-cybersec", "round": 9, "status": "ok", "context": 79401, "delta": 56622, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "floor", "round": 0, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "identical", "round": 0, "status": "ok", "context": 32731, "delta": 9952, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "unique", "round": 0, "status": "ok", "context": 33833, "delta": 11054, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "floor", "round": 1, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "identical", "round": 1, "status": "ok", "context": 32731, "delta": 9952, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "unique", "round": 1, "status": "ok", "context": 33849, "delta": 11070, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "floor", "round": 2, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "identical", "round": 2, "status": "ok", "context": 32731, "delta": 9952, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "unique", "round": 2, "status": "ok", "context": 33853, "delta": 11074, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "floor", "round": 3, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "identical", "round": 3, "status": "ok", "context": 32731, "delta": 9952, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "unique", "round": 3, "status": "ok", "context": 33867, "delta": 11088, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "floor", "round": 4, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "identical", "round": 4, "status": "ok", "context": 32731, "delta": 9952, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "unique", "round": 4, "status": "ok", "context": 33826, "delta": 11047, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "floor", "round": 5, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "identical", "round": 5, "status": "ok", "context": 32731, "delta": 9952, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "descuniq", "arm": "unique", "round": 5, "status": "ok", "context": 33887, "delta": 11108, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "floor", "round": 0, "status": "ok", "context": 22781, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "full", "round": 0, "status": "ok", "context": 40133, "delta": 17352, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "stripped", "round": 0, "status": "ok", "context": 79381, "delta": 56600, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "floor", "round": 1, "status": "ok", "context": 22781, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "full", "round": 1, "status": "ok", "context": 79373, "delta": 56592, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "stripped", "round": 1, "status": "ok", "context": 79381, "delta": 56600, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "floor", "round": 2, "status": "ok", "context": 22781, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "full", "round": 2, "status": "ok", "context": 79373, "delta": 56592, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "stripped", "round": 2, "status": "ok", "context": 79381, "delta": 56600, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "floor", "round": 3, "status": "ok", "context": 22781, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "full", "round": 3, "status": "ok", "context": 79373, "delta": 56592, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "stripped", "round": 3, "status": "ok", "context": 79381, "delta": 56600, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "floor", "round": 4, "status": "ok", "context": 22781, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "full", "round": 4, "status": "ok", "context": 40133, "delta": 17352, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "stripped", "round": 4, "status": "ok", "context": 79381, "delta": 56600, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "floor", "round": 5, "status": "ok", "context": 21884, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "fmcharge", "arm": "full", "round": 5, "status": "ok", "context": 79373, "delta": 57489, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "fmcharge", "arm": "stripped", "round": 5, "status": "ok", "context": 79381, "delta": 57497, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": true, "descs_match": true }, { "harness": "realdesc", "arm": "floor", "round": 0, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 0, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 0, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 1, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 1, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 1, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 2, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 2, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 2, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 3, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 3, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 3, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 4, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 4, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 4, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 5, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 5, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 5, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 6, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 6, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 6, "status": "ok", "context": 79381, "delta": 56602, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "floor", "round": 7, "status": "ok", "context": 22779, "delta": 0, "reply_exact_ok": true, "skills_on_disk": null, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "real_desc_only", "round": 7, "status": "failed", "context": 67351, "delta": 44572, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null }, { "harness": "realdesc", "arm": "full_pack", "round": 7, "status": "ok", "context": 79419, "delta": 56640, "reply_exact_ok": true, "skills_on_disk": 817, "names_match": null, "descs_match": null } ] }