Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add commands/intense-visions/harness-engineering/bug-fleetgit clone --depth 1 https://github.com/Intense-Visions/harness-engineeringWhat it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5 | $0.00141 | $0.00781 |
| Opus 5 | $0.00071 | $0.00391 |
| Sonnet 5 | $0.00028 | $0.00156 |
| Haiku 4.5 | $0.00014 | $0.00078 |
Grade A, and why
harness:bug-fleet scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
What it actually says
Phases:
- select: Enumerate disjoint, risk-ranked codebase areas by composing hotspot detection, critical-path and blast-radius analysis, a churn pass, and coverage depth as a risk multiplier, seeded by bugs parked by other quality-queue members
- confirm: Present the ranked areas, the pinned batch base SHA, the proposed concurrency, the bounded-safe vs risky-large fix boundary, the area-size bound, the candidates-per-area cap, and the reproduction-attempt budget for a single up-front human approval
- dispatch: Fan out worktree-isolated subagents, each running the real per-area hunt over standing code — review machinery, adversarial refutation, a tdd-authored reproducing test, tracker cross-check, fix classification, and a debugging-driven fix for bounded-safe defects only — capped at the concurrency governor
- verify: Independently confirm pipeline-provenance session artifacts exist, transplant each reproducing test onto the pinned base SHA and require an assertion-level failure, confirm it passes on the branch with the suite green, and confirm all-OS CI on every fix PR — never by subagent self-report
- file-and-report: Run the cross-area dedup backstop, open one fix PR per verified bug without merging, file each risky or large bug as an issue linking and quoting its reproduction branch, report security-routed findings to the human without publishing an exploit, and emit a one-row-per-item batch summary including discarded, refuted, already-known, and clean-area outcomes
<execution_context> @agents/skills/claude-code/bug-fleet/SKILL.md @agents/skills/claude-code/bug-fleet/skill.yaml </execution_context>
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 3d ago First seen · 41 lines · 141 tokens per session scan A c91e2499ef40
harness:bug-fleet is a command published in the GitHub repository Intense-Visions/harness-engineering (19 stars, last pushed 3d ago), licensed MIT. It adds 141 tokens to every session and 781 once invoked, about $0.0007 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
git
Git operations with intelligent commit messages and workflow optimization.
checklist
Generate a custom checklist for the current feature based on user requirements.
clarify
Identify underspecified areas in the current feature spec by asking up to 5 highly targeted clarification questions and encoding answers back into the spec.
specify
Create or update the feature specification from a natural language feature description.
analyze
Perform a non-destructive cross-artifact consistency and quality analysis across spec.md, plan.md, and tasks.md after task generation.
converge
Assess the current codebase against the feature's spec, plan, and tasks, then append any remaining unbuilt work as new tasks to tasks.md so implement can complete it.