mission-ledger
Skill14 starsChanged 7 months agoMITView on GitHub
The single source of truth for everything you've said yes to. If it's not in the ledger, it doesn't exist. If the ledger says it's done, it's done. mission-ledger owns the canonical task file: workspace/tasks/current.md. Every task that enters the autopilot's awareness — captured from email, from a voice note, from a call summary, from a direct owner prompt — lands here. The ledger is the authoritative state; no other skill maintains a parallel task list. workspace/tasks/current.md uses this exact…
In this repository7 other files it gives its agents.
- audience-echoSkill
- command-maestroSkill
- content-forgeSkill
- dawn-architectSkill
- knowledge-vaultSkill
- mail-sentinelSkill
- network-pulseSkill
The file
We split this skill into the rules it gives an agent
5 rules. About 0 are common, 3 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈3
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 48
Due date: when applicable — ISO 8601 only
- Very specific: line 92
Write HEARTBEAT_OK if nothing moved.
- Distinctive: line 98
When a new task is proposed (by the owner or by another skill), do not blindly append.
- Very specific: line 124
If more than 7 items sit in 🔥 Today, the skill refuses the 8th.
- Distinctive: line 132
Never: leave the file in an invalid markdown state (always atomic write → rename).
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
📋 Mission Ledger
The single source of truth for everything you've said yes to. If it's not in the ledger, it doesn't exist. If the ledger says it's done, it's done.
🎯 Mandate
mission-ledger owns the canonical task file: workspace/tasks/current.md. Every task that enters the autopilot's awareness — captured from email, from a voice note, from a call summary, from a direct owner prompt — lands here. The ledger is the authoritative state; no other skill maintains a parallel task list.
🗂️ The Canonical Format
workspace/tasks/current.md uses this exact structure. Deviating breaks other skills.
# Mission Ledger — <YYYY-MM-DD>
## 🔥 Today
- [ ] [P1] Review Q2 partnership draft — due 2026-04-18 — source: thread/abc123
- [ ] [P2] Record intro for module 7 — due 2026-04-18 — source: manual
## 📅 This Week
- [ ] [P2] Prepare deck for Paris AI Summit — due 2026-04-22 — source: cal/xyz789
## 🕐 Someday / Review
- [ ] [P3] Explore collab with Pr. Martin — source: thread/def456
## ✅ Completed Today
- [x] [P1] Ship v2.3 of Hermes Skills — completed 2026-04-18T14:30ZEvery task carries:
- Priority:
P1(today or the world ends),P2(this week),P3(nice-to-have) - Due date when applicable — ISO 8601 only
- Source pointer —
thread/<id>,cal/<id>,note/<id>,manual,dawn/<date>
🖐️ Interactive Operations
The owner talks to the ledger through five verbs:
add — capture a new task
> add: call Sarah about the book launch — P2 — due FridayAgent parses, normalizes the due date, and inserts into the correct section.
done — close a task
> done: review Q2 partnership draftAgent fuzzy-matches against current items, moves to ## ✅ Completed Today with a timestamp, and echoes what it closed so the owner can undo.
bump — change priority or date
> bump: Paris deck to P1
> bump: Sarah call to Mondaylist — show the current state
list today→ only the🔥 Todaysectionlist week→ Today + This Weeklist all→ full ledger
review — stand-up style summary
Three-line summary: open P1s, today's target, next blocker. Used by command-maestro for morning briefings.
🌙 Nightly Archive Pass (22:00)
Once a day, mission-ledger runs a janitorial pass:
- Move
## ✅ Completed Todayblock toworkspace/tasks/archive.mdunder a dated heading. - Promote overdue P2s → P1 and surface as an escalation only if the owner hasn't explicitly punted them today.
- Flag any P1 that has been open more than 72 hours to
command-maestrofor EOD review. - Write
HEARTBEAT_OKif nothing moved.
🧲 Magnetic Deduplication
When a new task is proposed (by the owner or by another skill), do not blindly append. Run:
- Exact-string match against open tasks → reject, echo the existing item.
- Semantic match (80%+ similarity on verb + object) → confirm with the caller once before adding.
- Otherwise → insert.
This prevents the ledger from being polluted by mail-sentinel when the same email arrives twice, or by the owner capturing a task they already wrote down yesterday.
🤝 Contract with Other Skills
| Skill | What it writes to the ledger | What it reads |
mail-sentinel | New tasks from actionable emails | Open tasks (dedup check) |
dawn-architect | Recurring tasks, calendar-implied prep | Yesterday's snapshot |
network-pulse | Follow-up reminders with dates | VIP-related open tasks |
command-maestro | Nothing — read-only | Full ledger for briefings |
knowledge-vault | Tasks extracted from meeting notes | Open tasks (link notes to tasks) |
mission-ledger is the only skill that writes completions. Other skills propose; only the ledger disposes.
🧘 Resistance to Priority Inflation
If more than 7 items sit in 🔥 Today, the skill refuses the 8th. It asks the owner to first close, bump, or kill something. This is deliberate — a list of 15 P1s is a list of zero P1s.
⚠️ Hard Never-Dos
- Never silently delete a task — always archive.
- Never re-prioritize without a matching
bumpfrom the owner or a nightly escalation. - Never leave the file in an invalid markdown state (always atomic write → rename).
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

