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.
git clone --depth 1 https://github.com/damusix/atomic-claudeWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/commands/damusix/atomic-claude/gather-evidence)<a href="https://agentmods.dev/commands/damusix/atomic-claude/gather-evidence"><img src="https://agentmods.dev/badge/commands/damusix/atomic-claude/gather-evidence.svg" alt="Measured on agentmods" height="20"></a>What 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 5.1 | $0.00074 | $0.02811 |
| Opus 5 | $0.00037 | $0.01406 |
| Sonnet 5 | $0.00015 | $0.00562 |
| Haiku 4.5 | $0.00007 | $0.00281 |
Grade A, and why
gather-evidence scanned grade A with 1 finding 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 8d 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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- **External service / API behavior** → `curl` with `-s -o /dev/null -w '%{http_code}'`, verify status and response shape. How it starts
The opening of the file, as written. The whole thing — 159 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/gather-evidence
Pre-design verification. The user has a hunch — "X library supports Y", "our codebase already has a Z pattern we can reuse", "approach A is faster than B" — and wants evidence before sinking a planning session into it. This command does that gathering and returns a structured verdict.
Not a code reviewer (atomic-reviewer). Not a strategist (atomic-strategist). Not a debugger (atomic-debug skill). Not a completion check (atomic-verify). One job: gather evidence on a hypothesis before it becomes a spec.
Workflow position
hunch → /gather-evidence → /pressure-test → /atomic-plan → /challenge-swarm → /subagent-implementation
(does it hold?) (is design right?) (write spec) (lenses attack it) (build it)
/gather-evidence is the first gate. Don't sink a planning session into a hunch that disintegrates on contact with reality.
Parse arguments
$ARGUMENTS may be empty, a hypothesis phrase, or a path to a markdown file.
- Token starts with
@or ends in.md(resolved relative to cwd, absolute allowed, must exist) → target document. Read it. Treat the document's central claims as the hypothesis set. - Anything else → hypothesis seed prose. Use it as-is.
- Empty
$ARGUMENTS→ scrape the recent conversation for the most recent unverified factual claim the user made (not your own assertions). If multiple candidates exist, pick the most recent one and ask the user to confirm or substitute before continuing. If none exists, ask: "What hypothesis do you want me to gather evidence on?"
Path safety. Resolve absolute path. If the file doesn't exist or isn't markdown, print one line: path '<x>' not found (or not markdown) — pass a hypothesis phrase instead. and stop. If the path resolves outside git rev-parse --show-toplevel, print: path '<x>' outside repo — continuing without target. and stop.
Four-step skeleton
1. RESTATE
Restate the hypothesis as a falsifiable claim. One sentence. Name what would have to be true.
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.
- 8d ago First seen · 159 lines · 74 tokens per session scan A dbf7ebdc7dd9
gather-evidence is a command published in the GitHub repository damusix/atomic-claude (85 stars, last pushed today), licensed MIT. It adds 74 tokens to every session and 2,811 once invoked, about $0.0004 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other commands, from other repositories
add-agent
Add a new sub-agent to the current plugin for specialized tasks.
maintain-check
Run one durable maintainer autopilot cycle — budget-gated, stop-guarded, checkpointed. Use --dry-run to preview without executing.
add-command
Add a new slash command to the current plugin.
add-skill
Add a new Skill to the current plugin with proper structure.
dream
Overnight autoresearch + maintainer sweep that surfaces a MORNING REPORT — suggested changes, issues raised with fixes, and kept improvements. Shadow-first; never auto-pushes. Runs on-demand or scheduled for off-hours.
save
Save current work state for next session. Creates/updates .planning/ files (CHECKPOINT.md, STATE.md, settings.json) so Heimdall resumes with full context. NOT a rewind — saves forward progress. Run before closing a session or at any milestone.