Grayscale/obelisk
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(memory): persistent memory layer with remember/recall CodeAct API

Browse Source 
Add a memories table (survives index rebuilds) for agent-written
  conclusions with provenance (session, message range, project). The agent
  writes markdown files via Write tool (user-approved), then registers
  them via a --remember CodeAct script with remember(). Recall via
  memories() in --query scripts, filtered by project/session/time.

  Separates query (read-only, assertReadOnlySql) from remember (write)
  execution contexts in runtime.mjs.
This commit is contained in:
tommy0103
2026-06-10 02:05:27 +08:00
parent dff88bbb37
commit 34f3a164ab
9 changed files with 364 additions and 29 deletions
Download Patch File Download Diff File Expand all files Collapse all files
references/retrieval-semantics.md
+13 -8
View File
@@ -13,7 +13,7 @@ Classify the user's request before choosing tools.
|-------------|--------------|------------|-------------|
| project name/path, session, cwd, file, time range | scope | `sessions()`, exact SQL on `project_path`, `sessionId`, `fileHistory()` | broad FTS |
| workflow, subagent, tool call, summary, edit | artifact | `workflows()`, `subagents()`, `summaries()`, `tool_calls`, `tool_results` | all-session search |
| concept, conclusion, design history, vague memory | semantic | `search()`, summaries, bounded facet sweep | session dumps |
| concept, conclusion, design history, vague memory | semantic | `memories({ query })`, `search()`, summaries, bounded facet sweep | session dumps |
One-shot retrieval is not all-shot retrieval. A query script may perform
multiple steps, but the first locator should be the narrowest semantic fit. If a
@@ -23,6 +23,7 @@ unless scoped evidence is insufficient and `query_plan` says why.
Project-like fields are distinct:
- `sessions.project`: stored Claude Code project slug.
- `memories.project`: stored project slug copied onto registered memory records.
- `sessions.project_path`: reconstructed absolute project path.
- `messages.cwd`: working directory at message time.
- helper `project`: SQL `LIKE` over `sessions.project`, not exact membership.
@@ -62,23 +63,27 @@ Use the database shape before asking the model to read text.
Ordering and context are semantic:
- `sessions()`, `summaries()`, `workflows()`, and `failures()` are newest first.
- `sessions()`, `memories()`, `summaries()`, `workflows()`, and `failures()` are newest first.
- `fileHistory()` is oldest first.
- `search().context` is temporal neighbors in one session, not causal context.
- `context(uuid)` and `trace(uuid)` are for parent-chain/causal expansion.
### Evidence Before Conclusion
Obelisk stores original structure, not precompiled claims. It has sessions,
messages, summaries, tool calls/results, files, subagents, workflows, parent
chains, and raw JSONL windows. It does not store "claim", "stance",
"contradiction", or "conclusion" entities.
Obelisk's raw session layer stores original structure, not precompiled claims:
sessions, messages, summaries, tool calls/results, files, subagents, workflows,
parent chains, and raw JSONL windows. The memory layer can store
human-approved markdown conclusions, but treat them as prior notes to compare
against raw evidence when correctness matters.
For semantic questions, build a task-local evidence view:
```js
{
query_plan: { mode, scope, facets, limits },
prior_memories: [
{ id, path, session_id, created_at, summary }
],
evidence: [
{ type, id, session_id, timestamp, facet, snippet }
],
@@ -86,8 +91,8 @@ For semantic questions, build a task-local evidence view:
}
```
Then synthesize the conclusion in the final answer. Do not pretend the evidence
view is a stored Obelisk entity.
Then synthesize the conclusion in the final answer. Do not pretend the raw
evidence view is itself a stored Obelisk entity.
## Text Search Semantics