SkillAgentSearch skills...

night-market-completion-integrity-campaign

Bind loop 'done' to unfakeable gates. Use to harden egregore/herald loops or promote completion_integrity. Not for QA gates; use night-market-validation-and-qa.

Install / Use

npx skills add athola/claude-night-market --skill night-market-completion-integrity-campaign

Installs into whichever agent you are using.

About this skill
📄

SKILL.md

Installable skill definition

Quality Score

91/100

Supported Platforms

Universal

Tags

Our assessment of night-market-completion-integrity-campaign

night-market-completion-integrity-campaign scores 91/100 on our quality scale, 1186th of 4,716 Development & Engineering skills we index (top 26%).

Its SKILL.md is 28 KB long, well organised into 21 sections with 15 code examples: a thorough specification that gives an agent plenty to work with.

It has 338 GitHub stars, a meaningful sign that others use it.

Substance
30/30
Structure
20/20
Description
15/15
Adoption
11/20
Freshness
15/15

Maintenance, license and trust

  • The repository was last updated about 2 months ago, so night-market-completion-integrity-campaign is actively maintained.
  • Our last check on 2026-11-20 found the source still online.
  • It is released under the MIT license, a permissive license that allows use, modification and commercial use with attribution.
  • Its trust signals score 100/100, with no cautions. These come from repository metadata, not a code audit — read the skill file before letting an agent act on it.

Safety scan

No issues found

Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful.

AI review by kimi-k2.7-code on 2026-10-07. Automated pattern scan on 2026-10-07. It catches known dangerous patterns, not every risk — read a skill before letting an agent act on it.

night-market-completion-integrity-campaign compared with similar skills

All 4 of these similar skills score higher than night-market-completion-integrity-campaign; compare them before choosing.

SkillScoreStarsUpdatedFormat
night-market-completion-integrity-campaign (this skill)by athola9133857d agoSKILL.md
taste-skillby Leonxlnx100118.9k11d agoCLAUDE.md
hyperframesby heygen-com10087.1ktodayCLAUDE.md
ai-job-searchby ai-job-search-dev10072.6ktodayCLAUDE.md
claude-howtoby luongnv8910047.5k5d agoCLAUDE.md

Frequently asked questions

How do I install night-market-completion-integrity-campaign?
Run npx skills add athola/claude-night-market --skill night-market-completion-integrity-campaign. The install tabs above show the steps for each supported agent.
Which AI agents does night-market-completion-integrity-campaign work with?
It is written for Universal, as a SKILL.md file. Other agents that read the same format can often use it too.
Is night-market-completion-integrity-campaign safe to use?
Our scan of the whole file found no instruction hijacking, hidden characters, credential access, data exfiltration or destructive commands. An AI review of the same text found nothing harmful. It is MIT-licensed and scores 100/100 on trust signals. Skills are instructions an agent will follow, so read the file before installing it and do not approve commands you do not understand.
Is night-market-completion-integrity-campaign still maintained?
The repository was last updated about 2 months ago, so night-market-completion-integrity-campaign is actively maintained.

name: night-market-completion-integrity-campaign description: Bind loop 'done' to unfakeable gates. Use to harden egregore/herald loops or promote completion_integrity. Not for QA gates; use night-market-validation-and-qa.

Night Market Completion-Integrity Campaign

This is an executable campaign, phased and decision-gated, against the hardest live problem in this repository: making "done" in autonomous loops mean something the agent cannot fake, and earning the promotion of completion_integrity from default-off to default-on. Every command below was run against the repo on 2026-07-02 (v1.9.15) unless marked candidate. Run the phases in order. Each phase ends at a gate with expected observations and branch instructions.

Definitions

| Term | Meaning | |------|---------| | Autonomous loop | A session that keeps working without a human turn: egregore's orchestrator (/egregore:summon), the externally installed ralph-wiggum loop, herald's auto-continue Stop hook | | Stop hook | A Claude Code hook fired when the agent tries to end its turn. It prints {"decision": "approve"} (allow stop) or {"decision": "block", "reason": ...} (keep working) | | Completion integrity | The property that a loop's "done" signal is bound to a verifier the agent does not control, instead of the agent's own say-so | | False stop | The loop halts while verifiable work remains, or accepts a claimed completion that a check would have rejected | | False continue | The loop keeps working (blocks the stop) on a turn that was genuinely finished | | Verdict | The egregore quality-gate outcome: pass, pass-with-warnings, or fix-required (computed after a 3-attempt auto-fix loop) |

Problem statement

Three assets exist today. None of them, alone, binds "done" to an unfakeable gate:

| Asset | What it does | Default | Key commits | |-------|--------------|---------|-------------| | plugins/egregore/scripts/config.py field PipelineConfig.completion_integrity | When true, a fix-required verdict counts as a step failure (the item cannot reach completed with blocking findings) and merge is held for human review regardless of auto_merge | False | 83281337 (gate), cd903cbf (raw-JSON opt-in test) | | plugins/herald/hooks/double_shot_latte.py (Stop hook) | Deterministic continue/stop judge over the last assistant message, with an opt-in LLM second shot for the single ambiguous outcome | Deterministic only | 268cff89 (timeout cap + gating) | | plugins/imbue/skills/proof-of-work/modules/verifier-integrity.md | The theory: a gate earns trust only if the spec is validated separately from the code and the check is proven able to fail | Prose guidance | 29081fda |

The load-bearing weakness, verified by reading commit 83281337: the completion_integrity flag is real code with a tested load path, but its enforcement lives in agent instructions (plugins/egregore/agents/orchestrator.md, the egregore:quality-gate skill, and skills/summon/modules/pipeline.md). No Python code path blocks a pipeline transition. The manifest that records item status (.egregore/manifest.json) is written by the same agent the gate is supposed to bind. The campaign exists to close that gap with evidence.

Evidence bar

Success is measured, never judged by eye. The standing rules are drawn from two 2026-07-01 research passes, whose evidence now lives in .claude/rules/prefer-invariants-over-fallbacks.md (harness-loop findings) and plugins/imbue/skills/proof-of-work/modules/verifier-integrity.md (verifier findings), plus the load-bearing claims inlined below:

  • State the numbers a hypothesis predicts BEFORE running the measurement. A threshold chosen after seeing the data is not a gate.
  • One mechanism must explain all observations, including the negative ones.
  • Never let the generator be its own judge: LLM self-verification is measurably unreliable and self-critique can degrade output, so verdicts come from an independent verifier (prover-verifier separation, arXiv 2402.08115; see plugins/imbue/skills/proof-of-work/modules/verifier-integrity.md).
  • A green check proves spec satisfaction, never correctness. Every gate you add must itself be proven able to go red (verifier-integrity Guard 2: mutation or revert test).

Campaign map

| Phase | Question it answers | Gate to pass | |-------|--------------------|--------------| | P0 | Do today's gate and judge suites pass as documented? | Exact pass counts reproduced | | P1 | What does the gate log when enabled on real work? | Opt-in path exercised on a bounded item, logs captured | | P2 | What are the false-stop and false-continue rates? | Pre-registered numbers met | | P3 | Can the loop satisfy the gate without doing the work? | Refutation attempts run, holes documented or closed | | P4 | Is the default flip earned? | Change control passed |

P0: baseline the gate and judge suites

Run every suite from inside its plugin directory. Root pytest sets norecursedirs = plugins/*, so running from the repo root silently collects nothing (or raises ImportPathMismatchError). Every cd in this skill is relative to the repo root: start each command block from there.

cd plugins/egregore
uv run pytest tests/ -q

Expected (2026-07-02): 477 passed in about 2 seconds, preceded by a coverage table.

cd plugins/egregore
uv run pytest tests/test_config.py tests/test_quality_gate.py -q

Expected: 27 passed. tests/test_config.py alone is 13 passed and includes the two completion-integrity guards: test_completion_integrity_opt_in_roundtrip and test_completion_integrity_loads_from_raw_json (the real user opt-in path, added in cd903cbf. It also guards the field against silent removal, because _filter_fields would drop the key).

cd plugins/herald
uv run pytest tests/ -q

Expected: 105 passed in under 2 seconds. This suite contains test_llm_timeout_fits_within_hook_timeout, which asserts LLM_TIMEOUT_SECONDS (8) is strictly below the Stop-hook timeout registered in plugins/herald/hooks/hooks.json (10).

cd plugins/imbue
uv run pytest tests/unit/skills/test_proof_of_work.py -q

Expected: 17 passed. Note: imbue's pytest addopts force coverage artifacts on every run. There is no dedicated test for the verifier-integrity module itself (verified 2026-07-02 by grepping plugins/imbue/tests/ for verifier-integrity): the module is prose, covered only by the skill-structure test above.

Smoke-test the herald judge directly. All three probes below were run and their outputs captured verbatim on 2026-07-02:

echo '{"session_id":"probe","transcript_path":"/nonexistent"}' \
  | python3 plugins/herald/hooks/double_shot_latte.py

Expected:

{"decision": "approve", "reason": "Double Shot Latte: No transcript available; allowing stop."}
D=$(mktemp -d)
printf '%s\n' '{"type":"assistant","message":{"role":"assistant","content":[{"type":"text","text":"Now let me fix the failing tests."}]}}' \
  > "$D/t.jsonl"
echo "{\"session_id\":\"p1\",\"transcript_path\":\"$D/t.jsonl\"}" \
  | python3 plugins/herald/hooks/double_shot_latte.py

Expected:

{"decision": "block", "reason": "Double Shot Latte: Assistant stated explicit intent to keep working."}

Gate P0 branches:

  • Pass counts lower than stated, or any failure: the repo has drifted since 2026-07-02. Stop the campaign. Triage with night-market-debugging-playbook, then update the counts in this skill's Provenance section before proceeding.
  • ImportPathMismatchError or no tests ran: you ran pytest from the wrong directory. Re-run from inside the plugin.
  • Hook probe emits nothing or a traceback: the hook contract is broken (it must always exit 0 and print a decision). That is a P0 incident, not a campaign step. File it.

P1: shadow observation, then opt-in on a bounded item

P1a: shadow read (no behavior change)

Verdicts are recorded in the manifest even with the flag off (the egregore:quality-gate skill records every verdict as a decision entry {"step": ..., "chose": ..., "why": ...}). So the flag-off world already contains the shadow data: items that advanced to completed despite a fix-required verdict. Count them in any repo where egregore has run:

python3 - <<'EOF'
import json
from pathlib import Path

path = Path(".egregore/manifest.json")
if not path.exists():
    print(json.dumps({"error": "no manifest; egregore has not run here"}))
    raise SystemExit(0)
data = json.loads(path.read_text())
items = data.get("work_items") or data.get("items") or []
flagged = [
    {"id": i.get("id"), "status": i.get("status")}
    for i in items
    if any(d.get("chose") == "fix-required" for d in i.get("decisions", []))
]
print(json.dumps(
    {"total_items": len(items), "items_with_fix_required": flagged},
    indent=2,
))
EOF

Expected: a JSON summary. Any entry with "status": "completed" in items_with_fix_required is a shadow-mode false completion: the exact event the gate exists to prevent. Record the count as the P2 baseline.

Branch: no manifest exists in this repo's root (egregore is typically summoned in target repos, and .egregore/ state lives where it ran). If you have no manifest anywhere, skip to P1b and generate one.

P1b: enable the gate on a bounded run

The opt-in path is hand-editing the config JSON. This exact shape is what test_completion_integrity_loads_from_raw_json covers: unspecified pipeline fields keep their defaults.

mkdir -p .egregore
cat > .egregore/config.json <<'EOF'
{"pipeline": {"completion_integrity": true}}
EOF

Warning: if .egregore/config.json already exists, edit the existing pipeline object instead of overwriting the file, or you will reset overseer and alert settings to defaults.

Then run a small, disposable work item in bounded mode (bounded mode stops when the time window expires, so the loop cannot run away while you observe):

/egregore:summon "<one small, well-specified task>" --bounded --window 5h

Before summoning, read "Stopping and relaunch machinery" at the end of this section: bounded mode expires on its own, but the watchdog and SessionStart hook can still resurrect the loop.

What it logs and where:

  • .egregore/manifest.json: per-item status (active, paused, pending count as unfinished for the Stop hook, while completed and failed are terminal), attempts, max_attempts (default 3), and the decisions array holding each quality verdict.
  • .egregore/relaunch-prompt.md: the re-injection prompt the egregore Stop hook (plugins/egregore/hooks/stop_hook.py) uses when it blocks an exit with active work remaining.
  • Overseer alerts per .egregore/config.json (pipeline_failure fires when an item exhausts attempts).

Expected observations with the flag on, per the documented contract in plugins/egregore/skills/quality-gate/SKILL.md and agents/orchestrator.md:

  1. A fix-required verdict routes to failure handling: the item retries in place, and after max_attempts it is marked failed with the overseer alerted. It is never silently completed.
  2. Merge is held for human review regardless of auto_merge: the PR is prepared but left open.
  3. The loop itself does not halt: it continues with the next active item.

Gate P1 branches:

  • An item reaches completed with an unresolved fix-required decision while the flag is true: the orchestrator ignored its instructions. This confirms the prompt-level enforcement gap in the problem statement. Record the manifest as evidence and carry it into P3. The finding argues for solution (a) or (b) below, which move enforcement out of the prompt.
  • The item loops in retry forever: max_attempts is not being incremented. That is an orchestrator bug, not a gate result. File it with the manifest attached.

Stopping and relaunch machinery

Know the stop path before you

Truncated for display — read the full file on GitHub.

Related Skills

View on GitHub
GitHub Stars338
CategoryDevelopment
Updated1mo ago
Forks35

Languages

Python

Trust signals

100/100

From repository metadata: license, adoption, age and documentation. Not a code audit — see the Safety scan above for what the skill file itself contains.

No cautions