synapse
Skill9 starsChanged 3 months agoMITView on GitHub
Use after a task completes to propagate learning across connected skills. Triggered by Cortex at the end of every successful task pipeline. Also triggers on "propagate learning", "update skills", "synapse", or when skills need cross-pollination after discovering shared patterns or errors.
10 other projects use a version of this.
Adopted by
Most recently taken up by
- thiagofinch/mega-braintheir own version, first seen Oct 8, 2026
- SynkraAI/aios-coretheir own version, first seen Oct 8, 2026
- anshulyadav1976/synapsetheir own version, first seen Sep 8, 2026
- yangjingo/Synapsetheir own version, first seen Aug 23, 2026
- hidrogenone/claude-synapsetheir own version, first seen Aug 12, 2026
- ANTHO56200/synapse-memorytheir own version, first seen Jul 6, 2026
Other versions10 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 10
- SynkraAI/aios-core★ 3.1ksame name
- SynkraAI/aiox-core★ 3.1ksame name
- thiagofinch/mega-brain★ 95same name
- anshulyadav1976/synapse★ 7same name
- ImScottDev/brain-scott★ 4same name
- yangjingo/Synapse★ 4same name
- NimbleBrainInc/skills★ 3same name
- ANTHO56200/synapse-memory★ 2same name
- TheJenilDGohel/Synapse★ 1same name
- hidrogenone/claude-synapse★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository6 other files it gives its agents.
- context-managerSkill
- cortexSkill
- design-firstSkill
- persona-labSkill
- pipeline-loggerSkill
- project-scaffolderSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. 0 are common, 8 are distinctive, and 7 are very specific to this project.
0
Common
Most of the 5 edited versions we read keep these.
8
Distinctive
This author's own choices. Worth a look.
7
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 12
Core principle: Every error found once should never recur — because Synapse propagates it to every skill that could encounter it.
- Very specific: line 51
When a new project is scaffolded (Phase 0), the scaffolder creates the initial synapse map by analyzing all generated skills and finding …
- Distinctive: line 67
Protected content rule: NEVER remove entries from "Anti-Patterns" or "Common Mistakes" during a refactor.
- Distinctive: line 69
Post-refactor validation: After rewriting, verify:
- Very specific: line 73
If validation fails → restore from backup → fall back to append-only mode (add [SYNAPSE] note)
- Very specific: line 77
Update last_synapse_update to today's date
- Very specific: line 83
Update connections list if new connections were discovered
- Distinctive: line 87
After updating the primary skill, evaluate: "Which other skills are impacted by this learning?"
- Very specific: line 107
After propagation, update skills/_synapse-map.md:
- Distinctive: line 108
Add/update connections for newly discovered relationships
- Distinctive: line 110
Add entry to "Recent Propagations" table
- Distinctive: line 114
If the learning includes an error or anti-pattern:
- Very specific: line 115
Add to guardrails/errors-to-avoid.md with:
- Distinctive: line 126
Remove connections referencing deleted skills
- Very specific: line 178
After each operation, log to logs/pipeline/YYYY-MM-DD.md under the SYNAPSE section of the current task:
“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Synapse — The Nervous System
Overview
Synapse makes skills learn from experience and communicate with each other. After every completed task, it updates the primary skill with new knowledge, propagates relevant insights to connected skills, and maintains the connection map that makes compound learning possible.
Core principle: Every error found once should never recur — because Synapse propagates it to every skill that could encounter it.
When Called
Synapse is called by Cortex at Step 8 of the task pipeline. It receives a "learning packet" containing:
- Task description and outcome
- Errors encountered during implementation
- Patterns that worked well
- QA feedback (what was rejected and why)
- Persona-Lab feedback (UX frictions found)
- Decisions made during the task
- Primary skill used
- Files changed
The Synapse Map
Maintained at skills/_synapse-map.md. This file tracks all skill relationships and propagation state.
Map format
# Synapse Map — Skill Connections
> Auto-generated and maintained by Synapse. Do not edit manually.
> Last updated: YYYY-MM-DD
## Connections
| Skill | Connected to | Connection type | Pending notes |
|-------|-------------|-----------------|---------------|
## Recent Propagations (last 20)
| Date | Origin | Learning | Propagated to |
|------|--------|----------|---------------|Map initialization
When a new project is scaffolded (Phase 0), the scaffolder creates the initial synapse map by analyzing all generated skills and finding shared concepts. See the project-scaffolder skill for the initialization logic.
The Four Operations
Operation 1: Update the primary skill (intelligent refactor)
Merge safety protocol:
- Backup before rewrite. Copy the current skill to
skills/[name]/.history/SKILL-YYYY-MM-DD-N.mdbefore any modification. This provides rollback capability. - Target canonical sections only. Integrate new knowledge into ONLY these sections:
- "Code Patterns" — add validated patterns
- "Anti-Patterns" / "Common Mistakes" — add newly discovered mistakes
- "Behavioral Rules" — reinforce or add rules
- "Verification Checklist" — add new checks
- Protected content rule. NEVER remove entries from "Anti-Patterns" or "Common Mistakes" during a refactor. These sections are append-only (entries may be reworded for clarity but never deleted).
- Post-refactor validation. After rewriting, verify:
- The refactored skill retains ALL anti-pattern entries from the backup
- The frontmatter is valid (name, description, maturity, connections)
- The skill is not shorter than 80% of the backup (guards against content loss)
- If validation fails → restore from backup → fall back to append-only mode (add
[SYNAPSE]note)
Metadata updates (Operation 1 responsibility):
- Increment
tasks_completedin frontmatter - Update
last_synapse_updateto today's date - Recalculate
maturity:- Seed: 0 tasks
- Growing: 1-3 tasks
- Mature: 4-9 tasks
- Expert: 10+ tasks
- Update
connectionslist if new connections were discovered
Operation 2: Map connections and propagate
After updating the primary skill, evaluate: "Which other skills are impacted by this learning?"
Propagation rules:
- Propagate ONLY to skills with direct connections in the synapse map (no cascading)
- Propagation is a contextual suggestion, not a rewrite — adds a tagged note:
[SYNAPSE from qs-android-widget, 2026-04-16] Caution with DateFormat
and IT locale — always use DateFormat(..., 'it_IT'). See errors-to-avoid.md.- Propagation counter: Each connection tracks
pending_notes. When a skill accumulates 5+ pending[SYNAPSE]notes, perform a full refactor of that skill (same merge safety protocol as Operation 1). Reset counter to 0 after refactor.
Connection detection heuristic (for new skills):
- Shared code patterns (same libraries, same data models)
- Shared domain concepts (dates, currency, user-facing text)
- Shared infrastructure (file I/O, notifications, database)
- Explicit references (one skill mentions another's domain)
Updating the synapse map:
After propagation, update skills/_synapse-map.md:
- Add/update connections for newly discovered relationships
- Increment
pending_notescounter for each propagated skill - Add entry to "Recent Propagations" table
Operation 3: Update global guardrails
If the learning includes an error or anti-pattern:
- Add to
guardrails/errors-to-avoid.mdwith:- Error ID (ERR-NNN, auto-incremented)
- Context: what was being done
- Error: what went wrong
- Fix: how to avoid it
- Skills affected: list of skill names
- Cross-reference the guardrail entry in the updated skill's anti-pattern section
Operation 4: Synapse map hygiene
Triggered at end of each session (called by the session-end checklist):
- Remove connections referencing deleted skills
- Trim "Recent Propagations" to last 20 entries
- Archive older entries to
skills/_synapse-history.md - Flag orphaned skills (connected to nothing) for user review
- Report: "Synapse map: N skills, M connections, K propagations this session"
Propagation Example
Task: "Implement Android home screen widget"
Specialist discovers: DateFormat crashes with IT locale on some Android versions
SYNAPSE executes:
1. BACKUP: skills/qs-android-widget/.history/SKILL-2026-04-16-1.md
2. UPDATE qs-android-widget:
- Add to "Common Mistakes":
| DateFormat IT locale crash | Use DateFormat('dd/MM/yyyy', 'it_IT')
with explicit locale, not platform default |
- Add code pattern: safe DateFormat initialization
- Increment tasks_completed: 0 → 1
- Update maturity: seed → growing
3. VALIDATE: anti-patterns retained? ✅ | frontmatter valid? ✅ | length >= 80%? ✅
4. CHECK connections in synapse map:
qs-android-widget ↔ qs-flutter-patterns (Widget lifecycle)
qs-android-widget ↔ qs-subscription-domain (Date logic) — DATE MATCH!
5. PROPAGATE to qs-subscription-domain:
[SYNAPSE from qs-android-widget, 2026-04-16] DateFormat with IT locale
can crash. Always pass explicit locale: DateFormat('dd/MM/yyyy', 'it_IT')
→ Increment pending_notes for this connection
6. PROPAGATE to qs-flutter-patterns:
[SYNAPSE from qs-android-widget, 2026-04-16] DateFormat locale safety —
always use explicit locale, never rely on platform default.
→ Increment pending_notes for this connection
7. UPDATE guardrails/errors-to-avoid.md:
## ERR-XXX: DateFormat IT locale crash
- Context: Android widget date display
- Error: DateFormat without explicit locale crashes on some devices
- Fix: Always pass 'it_IT' as second parameter
- Skills affected: qs-android-widget, qs-subscription-domain, qs-flutter-patterns
8. UPDATE skills/_synapse-map.md with new propagation entriesLogging
After each operation, log to logs/pipeline/YYYY-MM-DD.md under the SYNAPSE section of the current task:
- Operation 1 (Update skill): Log skill name, maturity change, patterns added, anti-patterns added
- Operation 2 (Propagate): Log each propagation: origin skill → target skill + what was propagated
- Operation 3 (Guardrails): Log ERR-NNN created, skills affected
- Operation 4 (Hygiene): Log summary: N skills, M connections, K propagations
On validation failure (backup restore): log to both pipeline/ and logs/errors/YYYY-MM-DD-errors.md with full context (what failed, what was restored).
See skills/pipeline-logger/SKILL.md for complete log format.
Anti-Patterns
| Don't | Do instead |
| Delete anti-pattern entries during refactor | Anti-patterns are append-only (reword, never delete) |
| Propagate to skills without direct connection | Only direct connections in synapse map |
| Skip the backup before rewriting | Always create .history/ backup first |
| Cascade propagation (A→B→C) | One hop only — B updates are B's responsibility next cycle |
| Refactor a skill without validation | Always check: entries retained, length >= 80%, frontmatter valid |
| Skip logging after propagation | Always log every propagation for audit trail |
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.

