Commitment Tracker

VerifiedCaution

Enforce immediate execution of any commitment made in a reply. Prevents the agent from saying it will do something without doing it.

Sby Skills Guide Bot
ProductivityIntermediate
7/22/2026
Claude CodeCursorCodex
#commitment-tracking#follow-through#agent-reliability#intent-log#crash-recovery

Recommended for

Automated summary

This skill forces the agent to carry out any action it promises in a reply during that same turn, instead of announcing it and silently dropping it.

Strengths

  • An explicit trigger-word list in both Hebrew and English (אדווח, I'll send, I'll follow up…) makes detection concrete rather than vague
  • A JSONL intent log written before the action and marked done afterwards, so unfulfilled commitments survive a gateway crash
  • A recovery scan designed for session start or heartbeat, plus cleanup of finished entries to keep the log near-empty in normal use
  • An honest fallback: when the action can't be performed, rewrite the reply so it promises nothing instead of leaving a dangling commitment

Limitations

  • The /path/to/workspace/data/commitments.jsonl path is a placeholder — without configuration the shell snippets target a nonexistent or wrong directory
  • Detection is purely literal keyword matching, so reworded promises ("consider it handled", "leave it with me") slip through unnoticed
  • The JSONL log has no locking or concurrency handling, and nothing confirms the action actually succeeded beyond writing a done record
When to use it

Use it on a conversational agent wired to outbound channels (DMs, group chats) that tends to promise actions without performing them in the same turn.

When not to use it

Skip it when the agent has no writable persistent workspace and no real ability to send messages, since both the intent log and immediate execution become impossible.

Security analysis

Caution
Quality score76/100

The skill is a behavioral guardrail against empty promises and contains no obfuscated or destructive payloads, and no tool permissions are declared. However it explicitly overrides confirmation norms by compelling the agent to perform real side-effecting actions (messaging, file updates) in the same turn and again during recovery scans, which is a meaningful autonomy risk even if intentional. Log handling uses normal shell file writes, so overall it lands at caution rather than risky.

Findings
  • •Instructs the agent to execute actions (send messages, update files, notify groups) autonomously and immediately without human confirmation, including on session-start/heartbeat recovery of 'pending' items — this can cause unintended outbound messages.
  • •Contains multiple embedded bash/python3 invocations that write to and rewrite a log file at a placeholder path (/path/to/workspace/data/commitments.jsonl); low risk but shell execution is required.
  • •Cleanup step truncates/rewrites the commitments log file in place, which could destroy audit history.
  • •No hardcoded secrets and no network exfiltration, curl|sh, base64 payloads, or safety-disabling content.
How this analysis is made

Examples

Follow up with the team
Check the deploy status and let the team know — I don't want another 'I'll update you later' that never happens.
Report back after checking logs
Look at the error logs and report back to me with what you find, actually done in this reply.
Wire up the commitment log
Set up data/commitments.jsonl in my workspace and make sure no promised action gets dropped between sessions.

name: commitment-tracker description: Enforce immediate execution of any commitment made in a reply. Use this skill before finalizing ANY reply that contains a promise to act — phrases like "אדווח", "אעדכן", "אשלח", "אחזור אליך", "I'll send", "I'll report", "I'll follow up", "I'll update". Prevents the agent from saying it will do something and then not doing it.

Commitment Tracker

The Rule

If your reply contains a commitment → execute it NOW, before sending the reply.

No deferred promises. No "I'll do it after". Either do it in this turn, or don't say it.

Commitment Trigger Words

Hebrew: אדווח, אעדכן, אשלח, אוסיף, אחזור אליך, אבדוק ואחזור, אתעדכן, אשאיר הודעה, אעביר

English: I'll send, I'll report, I'll update, I'll follow up, I'll check, I'll let you know, I'll message, I'll notify

Protocol

Before finalizing any reply:

  1. Scan your reply for trigger words above
  2. If found → execute the committed action now (send the message, update the file, etc.)
  3. Only after execution → include the result in your reply (e.g. "שלחתי ✅" not "אשלח")
  4. If you can't execute now → rewrite the reply to not promise it (say "לא יכולה לשלוח כרגע כי X")

Anti-Patterns

❌ "אדווח לנתנאל" → without actually messaging him
❌ "אעדכן את הקבוצה" → without actually sending to the group
❌ "אשלח לך סיכום" → without actually sending it

Correct Pattern

✅ Execute the action → then report: "עדכנתי את נתנאל ✅"
✅ If blocked: "לא יכולה לשלוח כי אין לי את ה-JID — צריכה עזרה"

Fault Tolerance — Intent Log

Before executing any commitment, write an intent record to prevent loss on gateway crash:

COMMITMENTS=/path/to/workspace/data/commitments.jsonl
mkdir -p $(dirname $COMMITMENTS)
echo "{\"ts\":\"$(date -u +%Y-%m-%dT%H:%M:%SZ)\",\"action\":\"DESCRIBE_ACTION\",\"target\":\"TARGET\",\"status\":\"pending\"}" >> $COMMITMENTS

After successful execution, mark done:

# Append a done record (simpler than editing in-place)
echo "{\"ts\":\"$(date -u +%Y-%m-%dT%H:%M:%SZ)\",\"action\":\"DESCRIBE_ACTION\",\"target\":\"TARGET\",\"status\":\"done\"}" >> $COMMITMENTS

Recovery — On Every Session Start / Heartbeat

Scan for unresolved commitments:

python3 -c "
import json
log = '/path/to/workspace/data/commitments.jsonl'
try:
    lines = [json.loads(l) for l in open(log)]
    done_actions = {l['action'] for l in lines if l.get('status')=='done'}
    pending = [l for l in lines if l.get('status')=='pending' and l['action'] not in done_actions]
    if pending:
        for p in pending: print(f\"PENDING: {p['action']} → {p['target']} (since {p['ts']})\")
except FileNotFoundError:
    pass
"

If any pending found → execute immediately and close the loop with the owner.

Cleanup — Delete Done Entries

After recovery scan, remove all done entries — they served their purpose:

python3 -c "
import json
log = '/path/to/workspace/data/commitments.jsonl'
try:
    lines = [json.loads(l) for l in open(log)]
    pending_only = [l for l in lines if l.get('status') != 'done']
    with open(log, 'w') as f:
        for l in pending_only: f.write(json.dumps(l)+'\\n')
except FileNotFoundError:
    pass
"

File only ever contains unresolved commitments — stays near-zero in normal operation.

Scope

Applies to ALL surfaces: DMs, groups, internal actions, cross-PA messages.
No exceptions for "it's obvious" or "they'll understand".

From the file to a first result

The download contains exactly the content displayed on this page. This copy may differ from the latest version in the source repository.

View the source and supporting files

This download contains SKILL.md only. Any scripts, references, templates and dependencies are not included. Get the required files from the source, preserving their directory structure, before continuing.

1. Download the file

Download SKILL.md

From the root of a test project, save the file as .skill-downloads/commitment-tracker-mrwakfso/SKILL.md, or use the command below. This folder is for inspecting the skill before installation.

Command for macOS/Linux or WSL: requires a POSIX shell and curl. On Windows without WSL, use the download button and create the folder manually.

mkdir -p '.skill-downloads/commitment-tracker-mrwakfso' && curl --fail --show-error --location 'https://skills-guides.com/api/skills/commitment-tracker-mrwakfso/download' --output '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md.part' && mv '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md.part' '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md'

2. Check prerequisites

Read SKILL.md before using it. If it references scripts or external services, inspect them and prepare the required access in your test environment.

No recognized prerequisites section in this file. This does not mean it is self-contained: libraries, tools and access requirements may be mentioned elsewhere in the content or source repository.

The platforms on this page are catalog indications. The paths below follow tool documentation; they do not prove this skill has been run in those tools. No minimum tool version is confirmed for this listing.

3. Load the skill

After inspection, copy the complete folder into your project using one of the commands below, then open a new tool session. The command stops if a folder with the same name already exists: compare the files before replacing it.

Claude Code
test -f '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md' && test -s '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md' && mkdir -p '.claude/skills' && test ! -e '.claude/skills/commitment-tracker' && cp -R '.skill-downloads/commitment-tracker-mrwakfso' '.claude/skills/commitment-tracker'

Cursor
test -f '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md' && test -s '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md' && mkdir -p '.cursor/skills' && test ! -e '.cursor/skills/commitment-tracker' && cp -R '.skill-downloads/commitment-tracker-mrwakfso' '.cursor/skills/commitment-tracker'

Codex
test -f '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md' && test -s '.skill-downloads/commitment-tracker-mrwakfso/SKILL.md' && mkdir -p '.agents/skills' && test ! -e '.agents/skills/commitment-tracker' && cp -R '.skill-downloads/commitment-tracker-mrwakfso' '.agents/skills/commitment-tracker'

OpenSkills

OpenSkills must already be installed. The local install/read procedure is checked with OpenSkills 1.5.0, which requires Node.js ≥ 20.6.0. Catalog slugs are not resolved automatically.

Official OpenSkills installation
openskills install './.skill-downloads/commitment-tracker-mrwakfso'

openskills read 'commitment-tracker-mrwakfso'

Reading should display the installed file content and its base directory. It does not run the skill or install its dependencies.

For an agent that reads AGENTS.md, then sync the index. This command changes AGENTS.md: review the diff before using it.

openskills sync

Manual method or another tool: attach SKILL.md to a conversation, or open it in an agent that can read files. Include any required supporting resources. No universal folder is recognized by every tool.

4. Try a first task

Use a short, synthetic input: an API endpoint, a small document or an interface brief. In a new conversation, attach the file or give the agent its path, then adapt this request:

Read .skill-downloads/commitment-tracker-mrwakfso/SKILL.md and apply its instructions to this example: [describe a test input]. Identify any missing files, dependencies or access. Produce the result requested by the skill without inventing data or running unreviewed scripts.

Expected result: output matching the skill instructions, or a precise list of missing prerequisites. A read confirmation alone does not validate a first use. If an external resource is unavailable, that path remains untested.

Related skills