@aabadin/project-memory-context 0.1.0

package/templates/opencode/autostart-snippet.md

@@ -0,0 +1,13 @@
+<!-- pmc:autostart -->
+When starting a session in a project with `.planning/project-memory-context/`:
+
+1. Check `.planning/project-memory-context/enrichment/worklist.json` for pending or stale entries.
+2. If pending/stale entries exist, launch background enrichment:
+   ```bash
+   {{PMC_BIN}} enrich .
+   ```
+3. Check `.planning/project-memory-context/enrichment/sync-manifest.json` for pending entries.
+4. If sync-manifest has pending entries, remind the user: "PMC has {n} pending sync operations. Run `/sync-context` to apply them."
+5. Search agent-memory for `project context overview` to fetch base context (stack, architecture, structure). Present a brief summary to establish session context.
+6. Remind the user: "Use `/get-context <target>` for structural deep-dive before reading files."
+<!-- /pmc:autostart -->

package/templates/opencode/commands/get-context.md

@@ -0,0 +1,22 @@
+---
+name: get-context
+description: Retrieve structured project context from agent-memory for the current project.
+argument-hint: "<target>"
+allowed-tools:
+  - Bash
+  - Read
+---
+
+<objective>
+Retrieve and display structured project context from agent-memory. Use before reading files to get a structural overview.
+</objective>
+
+<execution>
+Run:
+
+```bash
+{{PMC_BIN}} context . --refresh
+```
+
+This materializes or refreshes the 9 base project-context memories and outputs a summary.
+</execution>

package/templates/opencode/commands/new-project.md

@@ -0,0 +1,32 @@
+---
+name: new-project
+description: Bootstrap PMC in the current project with graphify, worklist, and base memories.
+argument-hint: "[--all] [--enrich]"
+allowed-tools:
+  - Bash
+  - Read
+  - Write
+  - Edit
+  - Glob
+  - Grep
+---
+
+<objective>
+Bootstrap PMC in the current project. Runs graphify, builds the enrichment worklist, materializes 9 base project-context memories, and writes the sync-manifest.
+</objective>
+
+<execution>
+Run:
+
+```bash
+{{PMC_BIN}} bootstrap . --all --enrich
+```
+
+Then follow up with `pmc status .` to verify the worklist was created.
+</execution>
+
+<success_criteria>
+- `.planning/project-memory-context/` directory exists with `graph/`, `enrichment/worklist.json`, and `enrichment/sync-manifest.json`
+- 9 base project-context memories materialized in `agent-memory`
+- Background enrichment started if `--enrich` was passed
+</success_criteria>

package/templates/opencode/commands/sanitize.md

@@ -0,0 +1,21 @@
+---
+name: sanitize
+description: Re-run graphify, diff symbols, mark stale entries, and optionally start background enrichment.
+argument-hint: ""
+allowed-tools:
+  - Bash
+---
+
+<objective>
+Sanitize the PMC enrichment state. Re-runs graphify, diffs the symbol graph, marks stale or removed entries, and optionally starts background enrichment.
+</objective>
+
+<execution>
+Run:
+
+```bash
+{{PMC_BIN}} sanitize .
+```
+
+This will re-run graphify, compare symbols against the existing worklist, mark stale entries, and delete orphaned enrichments. If pending items remain, background enrichment is launched automatically.
+</execution>

package/templates/opencode/commands/sync-context.md

@@ -0,0 +1,22 @@
+---
+name: sync-context
+description: Upsert enriched memories from the sync-manifest into agent-memory.
+argument-hint: ""
+allowed-tools:
+  - Bash
+---
+
+<objective>
+Process `.planning/project-memory-context/enrichment/sync-manifest.json` and upsert pending entries into `{{AGENT_MEMORY_CMD}}`.
+</objective>
+
+<execution>
+Run the sync using `agent-memory` MCP tools. For each pending entry in the sync-manifest, call `agent-memory_store` or `agent-memory_update` with the enrichment payload.
+
+After processing, mark entries as `synced` in the sync-manifest.
+</execution>
+
+<success_criteria>
+- All pending entries in sync-manifest marked as `synced`
+- No orphaned memories remaining
+</success_criteria>

package/templates/project-memory-context workflow.md

@@ -0,0 +1,129 @@
+<purpose>
+Run a resumable project-memory-context workflow for existing repositories.
+
+Stage A collects project intent, uses `brainstorming` to clarify ambiguity, persists intake artifacts, and invokes `graphify` as the structural mapping phase.
+
+Stage B scans supported files (`JS/TS + .NET`), extracts top-level symbols, computes stable symbol identities, and writes `.planning/project-memory-context/enrichment/worklist.json` for semantic enrichment and agent-memory upserts.
+</purpose>
+
+<process>
+
+<step name="parse_stage" priority="first">
+Parse `$ARGUMENTS` and normalize one of:
+
+- `stage-a`
+- `stage-b`
+- `all`
+
+Default to `stage-a` when omitted.
+</step>
+
+<step name="stage_a_intake" condition="stage-a or all">
+Ask the user for:
+
+- a short project description
+- the mapping goals
+- optional focus areas
+
+Then invoke the `brainstorming` skill to clarify any ambiguity in the intake before continuing.
+
+Persist the normalized intake using:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/save-intake-context.mjs" "<project description>" <goal1> <goal2>
+```
+</step>
+
+<step name="stage_a_graphify" condition="stage-a or all">
+Run `graphify` against the current workspace after intake has been clarified.
+
+Expected graph artifacts:
+
+- `.planning/project-memory-context/graph/graph.json`
+- `.planning/project-memory-context/graph/graph.html`
+- `.planning/project-memory-context/graph/GRAPH_REPORT.md`
+- `.planning/project-memory-context/graph/graph.metadata.json`
+
+If `graphify` writes to `graphify-out/`, copy or mirror the resulting graph artifacts into `.planning/project-memory-context/graph/`.
+</step>
+
+<step name="stage_b_scan" condition="stage-b or all">
+Select supported files from the repo for v1:
+
+- `*.js`, `*.jsx`, `*.ts`, `*.tsx`
+- `*.mjs`, `*.cjs`
+- `*.cs`
+
+Exclude generated folders such as `node_modules`, `dist`, `.git`, `bin`, `obj`.
+
+Build the deterministic worklist with:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/build-worklist.mjs" <relative-file-paths>
+```
+
+The script writes `.planning/project-memory-context/enrichment/worklist.json`.
+</step>
+
+<step name="stage_b_semantic_enrichment" condition="stage-b or all">
+Prepare semantic jobs with:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/prepare-semantic-jobs.mjs"
+```
+
+The script writes `.planning/project-memory-context/enrichment/semantic-jobs.json`.
+
+For each pending job in `semantic-jobs.json`:
+
+1. load the exact file fragment for the symbol range
+2. call the `pmc-local-model` MCP with the prepared prompt and require a `semantic_report` result using this stable prefix format:
+
+```text
+responsibility: ...
+inputs: item1, item2
+output: ...
+dependencies: dep1, dep2
+role: ...
+```
+
+3. materialize the memory payload before calling `pmc-agent-memory`:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/materialize-enrichment-artifacts.mjs" @<job-file> @<report-file>
+```
+
+4. persist or update the semantic payload in `pmc-agent-memory`
+
+5. once `pmc-agent-memory` returns the `memoryId`, materialize the graph update payload:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/materialize-enrichment-artifacts.mjs" @<job-file> @<report-file> <memory-id>
+```
+
+6. finalize the successful symbol using:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/finalize-enrichment.mjs" @<symbol>.result.json
+```
+
+If a symbol cannot be processed, mark it as `error` without aborting the rest of the run:
+
+```bash
+node "__PMC_PACKAGE_ROOT__/cli/fail-enrichment.mjs" <symbol-key> "<error message>"
+```
+</step>
+
+<step name="report">
+Print a compact summary:
+
+- intake saved
+- graph artifacts present or missing
+- supported symbols found
+- worklist count
+- pending / enriched counts when Stage B runs
+
+End by pointing to `.planning/project-memory-context/` as the persistent run directory.
+</step>
+
+</process>

package/templates/project-memory-context.md

@@ -0,0 +1,42 @@
+---
+name: project-memory-context
+description: Intake an existing project, clarify mapping goals, run graphify, and build a resumable semantic enrichment worklist for JS/TS and .NET symbols.
+argument-hint: "[stage-a|stage-b|all] [--paths <p1,p2,...>]"
+allowed-tools:
+  - Read
+  - Bash
+  - Glob
+  - Grep
+  - Edit
+  - Write
+  - Agent
+  - TaskOutput
+  - Question
+  - Skill
+---
+
+<objective>
+Create a persistent project memory context for an existing codebase by combining intake context, graphify structural mapping, symbol extraction, local semantic enrichment, and agent-memory persistence.
+</objective>
+
+<execution_context>
+@./project-memory-context workflow.md
+</execution_context>
+
+<context>
+Arguments: $ARGUMENTS
+
+Supported stages:
+- `stage-a` — intake, clarification, graphify, artifact persistence
+- `stage-b` — load graph artifacts, build symbol worklist, prepare semantic enrichment
+- `all` — run Stage A then Stage B
+
+Default stage is `stage-a`.
+</context>
+
+<success_criteria>
+- Intake artifacts persisted under `.planning/project-memory-context/intake/`
+- Graph artifacts persisted under `.planning/project-memory-context/graph/`
+- Enrichment worklist persisted under `.planning/project-memory-context/enrichment/worklist.json`
+- Workflow leaves resumable metadata for future semantic enrichment runs
+</success_criteria>