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 skills add suyoumo/ClawProBench --skill commitment-setupgit clone --depth 1 https://github.com/suyoumo/ClawProBenchWrote 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/skills/suyoumo/clawprobench/commitment-setup)<a href="https://agentmods.dev/skills/suyoumo/clawprobench/commitment-setup"><img src="https://agentmods.dev/badge/skills/suyoumo/clawprobench/commitment-setup/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/suyoumo/clawprobench/commitment-setup"><img src="https://agentmods.dev/badge/skills/suyoumo/clawprobench/commitment-setup.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00058 | $0.02602 |
| Opus 5.5 | $0.00023 | $0.01041 |
| Sonnet 5.5 | $0.00012 | $0.00520 |
| Haiku 4.5 | $0.00006 | $0.00260 |
Grade A, and why
commitment-setup 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 1mo 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 — 251 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Commitment System Setup
You are installing the commitments tracking system. This creates a workspace structure for tracking obligations, signals, decisions, and parked ideas, plus two missions for automated triage and digest delivery.
Step 1: Check existing setup
Call memory_read(path="projects/commitments/README.md"). If it exists, tell the user: "The commitments system is already set up. Want me to reinstall from scratch?" Stop unless they confirm.
Step 2: Gather user context
The user's timezone is provided by the channel automatically — do not ask for it. Ask only:
- Which channel should I send digests to? (default: the current channel)
Step 3: Write the schema README
Call memory_write with target="projects/commitments/README.md", append=false, and this content:
# Commitments System
Tracks obligations, decisions, and ideas via structured markdown files.
## Directory Layout
- `open/` — Active commitments (one file each)
- `resolved/` — Completed commitments (archived)
- `signals/pending/` — Raw extracted signals awaiting triage
- `signals/expired/` — Signals that were not promoted in time
- `decisions/` — Captured decisions with rationale
- `parked-ideas/` — Ideas saved for later consideration
## Signal Schema (signals/pending/<slug>.md)
---
type: signal
source_channel: <channel name>
source_message: "<brief quote or paraphrase>"
detected_at: <YYYY-MM-DD>
immediacy: realtime | prompt | batch
expires_at: <YYYY-MM-DD> | null
confidence: high | medium | low
obligation_type: reply | deliver | attend | review | decide | follow-up | informational
mentions: [<names>]
destination: null | commitment | parked_idea | intelligence | dismissed
promoted_to: null | <commitment filename>
---
<Human-readable description of what was detected.>
### Immediacy levels
- realtime: push-notify immediately (production incident, market alert, security)
- prompt: surface within the hour (urgent DM, trending topic)
- batch: next digest is fine (meeting action item, report to read)
### Signal destinations
- commitment: actionable, tracked, has a resolution path
- parked_idea: interesting, not committed, revisit later
- intelligence: no action needed, but informs future decisions → write a MemoryDoc via memory_write
- dismissed: not relevant
## Commitment Schema (open/<slug>.md or resolved/<slug>.md)
---
type: commitment
status: open | in_progress | blocked | waiting | resolved
urgency: critical | high | medium | low
due: <YYYY-MM-DD> | null
created_at: <YYYY-MM-DD>
stale_after: <YYYY-MM-DD> | null
owner: user | agent
delegated_to: null | <person or team>
resolution_path: agent_can_handle | needs_reply | needs_decision | note_only
decision_type: mechanical | taste | challenge
effort_human: <estimate> | null
effort_assisted: <estimate> | null
source_signal: <relative path> | null
resolved_by: null | agent | user | delegate | expired
tags: [<freeform>]
---
# <Title>
<Description of the obligation.>
## Resolution path
- [ ] Step 1
- [ ] Step 2
## Progress
<Updates appended over time.>
### Resolution path types
- agent_can_handle: the agent can do this autonomously (review PR, draft doc, research)
- needs_reply: user must send a response to someone
- needs_decision: user must choose between options
- note_only: informational, no action needed but tracked
### Decision types (when to ask the user)
- mechanical: auto-act silently (expire stale signal, update status, dismiss noise). Report in digest as "auto-handled."
- taste: auto-act but surface for awareness ("I auto-dismissed 3 FYI signals, auto-resolved 2 completed items"). User can override.
- challenge: always ask the user before acting (architecture decisions, sending messages, spending money, irreversible actions).
### Effort estimates
When known, include dual estimates:
- effort_human: time without AI assistance (e.g. "2h", "3d")
- effort_assisted: time with AI assistance (e.g. "15min", "2h")
This reframes decisions — when AI makes completeness cheap, there is no excuse for shortcuts.
### Autonomous resolution
When a commitment has resolution_path=agent_can_handle and decision_type=mechanical or taste:
1. Agent handles it and reports in the next digest
When decision_type=challenge:
1. Agent asks for explicit approval before acting
2. On approval, spawns a mission for complex work or handles inline
3. Status transitions to in_progress, then resolved with resolved_by=agent
## Decision Schema (decisions/<date>-<slug>.md)
---
type: decision
decided_at: <YYYY-MM-DD>
context: <topic slug>
participants: [<names>]
confidence: high | medium | low
reversible: true | false
outcome: null | <brief outcome description>
outcome_positive: null | true | false
tags: [<freeform>]
---
# <What was decided>
## Context
<Why this decision was needed.>
## Options considered
1. **Option A** — pros/cons
2. **Option B** — pros/cons
## Rationale
<Why this option was chosen.>
## Outcome
<Filled in later: what happened as a result of this decision.>
## Parked Idea Schema (parked-ideas/<slug>.md)
---
type: parked-idea
parked_at: <YYYY-MM-DD>
source: conversation | triage | research
relevance: high | medium | low
tags: [<freeform>]
---
# <Idea title>
<Description and why it is interesting.>
## Activation trigger
<What would make this worth pursuing.>
## Conventions
- Filenames use lowercase kebab-case: `review-sarah-deck.md`
- Dates are ISO-8601: `YYYY-MM-DD`
- Moving a commitment from open/ to resolved/: write the updated file to resolved/, then overwrite the open/ file with empty content
- One file per entity — never batch multiple commitments into one file
## Trust calibration
Start conservative:
- All signals surfaced in digest, none auto-promoted to commitments
- All agent_can_handle commitments require explicit user approval before dispatch
- Realtime immediacy disabled initially (everything batched)
- Track user feedback patterns in projects/commitments/calibration.md to gradually increase autonomy
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.
- 1mo ago First seen · 251 lines · 58 tokens per session scan A 3dd6177f1b53
commitment-setup is a skill published in the GitHub repository suyoumo/ClawProBench (827 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 58 tokens to every session and 2,602 once invoked, about $0.0002 per session on Opus 5.5. 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
company-employee
The protocol every employee of a PenguinHarness organization follows — read the handbook first, act on [orgtrigger] work runs, schedule tickets from the desk session and do the work in ticket sessions, block instead of idling, ask the board in the all-hands channel before anything heavy, costly, irreversible or…
company-ceo
Run a PenguinHarness organization as its CEO — turn the mission into a ticket tree, hire HR and finance first, partition the shared workspace, schedule the calendar, open a channel per stream, review tickets and report to the board in the all-hands channel.
company-hr
Run HR for a PenguinHarness organization — guarantee every employee has an enabled calendar event, hire and offboard employees (channels included), evaluate and improve them (with the default-rsi-toolkit plugin's agent-optimization skill), and keep the handbook's role conventions current.
company-mirror
Run a PenguinHarness organization that mirrors a real company — one digital twin per real colleague, each twin's desk bound to that colleague's chat bot, twins answering what the handbook already knows and relaying the rest to one another instead of hiring, scheduling and filing tickets.
company-setup
Create a company-mode organization together with the user — collect its id, name, mission, shared workspace, model and CEO budget one question at a time, confirm the whole thing in one summary, then run penguin org create and hand the user over to company mode.
caixun-work-hours
A work-hours reporting assistant for RichPMS, a project time-tracking system. It turns conversation details, code records, or your description into a daily report, lets you confirm it, and then saves or submits it.