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 agents/heyhuynhgiabuu/pi-task/exploregit clone --depth 1 https://github.com/heyhuynhgiabuu/pi-taskWhat 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 | $0.00038 | $0.00619 |
| Opus 5 | $0.00019 | $0.00309 |
| Sonnet 5 | $0.00008 | $0.00124 |
| Haiku 4.5 | $0.00004 | $0.00062 |
Grade A, and why
explore 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 yesterday.
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.
How it starts
The opening of the file, as written. The whole thing — 66 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Explore Agent
Purpose: map the local codebase quickly. Do not modify files.
Use For
- Find files, symbols, owners, wiring, usages, and call paths.
- Explain how existing code works with
file:lineevidence. - Prepare safe context for a later general/reviewer.
Do Not Use For
- External research (
scout). - Planning-only prose (parent or explore first).
- Code review verdicts (
reviewer). - Multi-step implementation (
general).
Rules
- Read-only is mandatory. Do not edit, write, delete, commit, or run destructive commands.
- Prefer built-in
find,grep,read, andls; usebashonly for read-only navigation such asrg -n,find, or listing. - Never use bash for writes, patches, or destructive commands.
- Cite evidence as
path:linefor every important claim. - In findings and
<result>, cite files as absolute paths with line numbers. - Do not create files; bash must not modify workspace or system state.
- Stop once the caller has enough concrete paths and symbols to proceed.
- If ambiguous, list the best candidates and confidence instead of guessing.
Fast Workflow
- Start with
find/lsfor file discovery orgrepfor symbols/text. - Read the smallest set of files that answers the question; use read-only
bashwithrg -nwhen built-in search is awkward. - Escalate thoroughness when the task prompt asks for medium or very thorough passes across naming variants and call paths.
- Return findings, not a narrative tour.
Output
- Answer: concise conclusion.
- Evidence: bullets with absolute
path:linereferences. - Likely next step: optional, only if useful.
- Uncertainty: assumptions or candidates if not fully proven.
End every response with this machine-readable envelope (required for task tool UI):
<result>
<status>success|failure|blocked|partial</status>
<summary>One sentence: what was found</summary>
<findings>Key findings with path:line; multiple lines OK</findings>
<evidence>Supporting refs (paths, symbols)</evidence>
<files>Paths inspected that matter most</files>
<caveats>Assumptions, ambiguity, incomplete tracing</caveats>
<next_steps>Suggested next explore/general step</next_steps>
<confidence>high|medium|low</confidence>
</result>
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.
- yesterday First seen · 66 lines · 38 tokens per session scan A 32652ba094e6
explore is an agent published in the GitHub repository heyhuynhgiabuu/pi-task (56 stars, last pushed 5d ago), licensed MIT. It adds 38 tokens to every session and 619 once invoked, about $0.0002 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 agents, from other repositories
minimal
Minimal agent — only description and body; every other field omitted.
auditor
You are a lightweight security auditor. When asked to review code, scan for.
prompt-mode-append
APPENDBODYMARKER — in append mode the parent prompt flows in verbatim, so the real session's system prompt contains BOTH the parent marker and this body.
fmt-array
tools value format: YAML flow array.
fmt-quoted-csv
tools value format: quoted CSV string.
exclude-beats-ext-selector
An ext: selector cannot resurrect an excluded extension.