feat: release v0.2 self-evolving layer — skills, memory graph, reflections#9
Open
mayankbohradev wants to merge 1 commit into
Open
feat: release v0.2 self-evolving layer — skills, memory graph, reflections#9mayankbohradev wants to merge 1 commit into
mayankbohradev wants to merge 1 commit into
Conversation
…reflections Adds Hermes-style procedural memory + LLM Wiki-style typed link graph on top of the existing flat memory store. Zero breaking changes — every existing tool keeps its signature, every existing memory keeps its data. Schema (additive migration 0002_self_evolving.sql): - memories gains 6 columns: confidence, verified_at, superseded_by, tier, access_count, last_accessed_at (all with safe defaults) - New tables: skills + skills_fts, memory_links, reflections, injection_log Worker (src/index.ts): - 10 new MCP tools: save_skill, get_skill, list_skills, record_skill_outcome, link_memories, get_memory_graph, list_reflections, resolve_reflection, get_hub_health, verify_memory - Injection scanner with 8 rules on every write surface - Auto-detection of contradictions (similarity 0.3-0.6 + opposing signal words) - Auto-detection of skill candidates (procedural markers in learning/decision) - Nudge layer: every save/search appends top-3 pending reflections CLI (create-context-hub@0.4.2): - update command always probes remote D1 for 0.2 schema before deciding "already up to date" — fixes the source-repo edge case - Auto-applies 0002 migration if D1 still on 0.1 schema (with confirmation) - Plain-English intro explaining what update does + what is preserved - Smarter redeploy gate — offers redeploy when DB migrates even if files match - Post-migration re-probe confirms the schema actually landed
|
@mayankbohradev is attempting to deploy a commit to the Jaipuria AI Labs Team on Vercel. A member of the Team first needs to authorize it. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Context Hub 0.2 — the self-evolving layer. Hermes-style procedural memory + LLM Wiki-style typed link graph on top of the existing flat memory store.
Zero breaking changes. Every existing tool keeps its signature, every existing memory keeps its data.
What's in this release
Schema (additive migration
0002_self_evolving.sql):memoriesgains 6 columns:confidence,verified_at,superseded_by,tier,access_count,last_accessed_at(all with safe defaults)skills+skills_fts,memory_links,reflections,injection_logWorker (
src/index.ts) — 10 new MCP tools:save_skill,get_skill,list_skills,record_skill_outcome— procedural memorylink_memories,get_memory_graph— typed graph (supersedes, contradicts, supports, related, example_of, refines)list_reflections,resolve_reflection— self-improvement queueget_hub_health,verify_memory— hygiene + decayPlus injection scanner (8 rules), contradiction auto-detection at write time, skill-candidate auto-flagging, and a nudge layer that appends top-3 pending reflections to every save/search response.
CLI (
create-context-hub@0.4.2):updatecommand always probes remote D1 for 0.2 schema (fixes source-repo edge case)Upgrade path for existing users
Existing memories are preserved with sensible defaults (
confidence=0.8,tier='warm'). No data migration.Test plan
npm run typecheckpasses on both root and CLI packageupdateflow tested on the source repo — probes correctly, migrates DB, re-probes to confirm