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 skills/suyoumo/clawprobench/commitment-triagenpx skills add suyoumo/ClawProBench --skill commitment-triagegit clone --depth 1 https://github.com/suyoumo/ClawProBenchWhat 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 | $0.00029 | $0.02206 |
| Opus 5 | $0.00015 | $0.01103 |
| Sonnet 5 | $0.00006 | $0.00441 |
| Haiku 4.5 | $0.00003 | $0.00221 |
Grade A, and why
commitment-triage 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 2d 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.
How it starts
The opening of the file, as written. The whole thing — 212 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Commitment Triage
You have a commitments tracking system in the workspace under projects/commitments/. Read projects/commitments/README.md for the full schema if you need field details.
Mode A: Passive signal detection
When the user says something that implies an obligation, promise, or deadline — but is NOT explicitly asking you to track it — silently extract a signal.
Triggers: "I need to...", "I promised Sarah...", "I should get back to...", "The report is due Friday", "They asked me to review..."
Treat these as especially strong passive-signal cases even if the user does not say "track this":
- "The strategy team asked me to review the international expansion reforecast this week."
- "Legal asked me to comment on the draft by Thursday."
- "Slack message from Priya: can you review the OAuth callback edge case this week?"
- "I need to review Sarah's deck before Friday."
These should usually become review, reply, or follow-up signals with
immediacy: prompt when the request is time-sensitive or comes from a named
person/team.
Treat inbound-message phrasings like these as strong passive signals too:
- "Slack message from strategy: can you review the international expansion reforecast this week?"
- "Email from legal: can you comment on the draft by Thursday?"
Action:
- Check for duplicates:
memory_searchfor key phrases withinprojects/commitments/ - If no duplicate, call
memory_writewith:target:projects/commitments/signals/pending/<slug>.mdappend: false- Content: signal frontmatter + description
- Only after the
memory_writesucceeds, at a natural pause briefly note: "I've tracked a commitment about [topic]."
Do not merely acknowledge or summarize an obligation. This mode is successful
only if a signal is actually written to projects/commitments/signals/pending/.
Do NOT interrupt the conversation flow. Signal extraction is a side-effect.
For commitment tracking, use memory_tree, memory_read, and memory_write.
Do not use CodeAct, shell commands, or creative-generation tools unless the
user explicitly asked for execution rather than tracking.
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.
- 2d ago First seen · 212 lines · 29 tokens per session scan A 3b1f8411a0a3
commitment-triage is a skill published in the GitHub repository suyoumo/ClawProBench (823 stars, last pushed 8d ago), licensed Apache-2.0. It adds 29 tokens to every session and 2,206 once invoked, about $0.0001 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 skills, from other repositories
lastlight-evals
Scaffold, configure and run a Last Light EVALS workspace — the harness that runs Last Light's real workflows against a mocked GitHub and grades them deterministically. Use when the user wants to "set up / scaffold Last Light Evals", "create an evals workspace or instance", "run evals", "compare models", or author new…
issue-comment
Handle a non-build maintainer comment on an issue or PR — close, reopen, label, dedupe, answer a brief question, or triage. Action-only; redirect anything that needs code changes to /build.
chat
Conversational assistant for messaging-platform threads (Slack, Discord). Answer questions about repos, PRs, and issues, explain code, and guide users to the natural-language workflow triggers listed in the system prompt.
Evaluation
Frames model, prompt, and system evaluation as a reproducible experiment with baselines, datasets, and explicit metrics.
perpetuum
为一个或多个项目建立、运行和管理长期 Agent 队伍。适用于初始化长期自主工作、按项目 cron 计划持续推进 Story、查看或调整 Story 看板，以及处理人类输入。.
openspec-archive-change
Archive a completed change in the experimental workflow. Use when the user wants to finalize and archive a change after implementation is complete.