docs: define milestone v1.1 requirements

This commit is contained in:
2026-06-14 18:06:39 +08:00
parent 4b58964a12
commit 719fb1e07a
2 changed files with 78 additions and 6 deletions

71
.planning/REQUIREMENTS.md Normal file
View File

@@ -0,0 +1,71 @@
# Milestone v1.1 Requirements
**Milestone:** v1.1 — Session Lifecycle, Memory & Reporting
**Core Value:** Agent must NEVER mutate real infrastructure beyond what the limited IAM role permits
**Target:** Productionize session workspace, persistent memory, and operational reporting
## Requirements
### Memory (MEM)
- [ ] **MEM-01**: Hindsight memory provider enabled for cross-session entity-aware recall
- Details: Switch from built-in MEMORY.md/USER.md to Hindsight Cloud API
- Config: `memory.provider: hindsight`, `HINDSIGHT_API_KEY` in .env
- Verification: Agent recalls facts across sessions without user prompting
### Repository Workspace (REPO)
- [ ] **REPO-01**: DEFAULT_REPOS (`rai-ops`, `rai-deployment`, `rai-devtools`) auto-cloned into every new Hermes session
- Details: `shell_init_files` script clones to host-mounted `/workspace/repos/` volume
- Config: `terminal.shell_init_files`, `terminal.docker_volumes` (SSH mount + workspace volume)
- Verification: New session has repos available at `/workspace/repos/`
- [ ] **REPO-02**: User can request additional repos to clone on demand during a session
- Details: Agent can clone arbitrary repos via git command in Docker
- Verification: User says "clone rai-pipeline", agent clones it
### Main Session Skill (SKIL)
- [ ] **SKIL-04**: Main ngn-agent session orchestration skill covering the initial-plan.md workflow
- Details: Hermes SKILL.md defining session lifecycle — detect similar prev sessions (via hindsight), load DEFAULT_REPOS, create Jira ticket per session, load Confluence docs, update docs/Jira at session end
- Verification: Agent follows the full session workflow on a new task
### Cron Reporting (CRON)
- [ ] **CRON-01**: Daily session summary report delivered via Telegram at 09:00
- Details: Hermes skill-backed cron job. Lists active sessions, last activity, token usage
- Verification: Telegram receives daily report
- [ ] **CRON-02**: Stale session auto-archive (30d inactivity) — export to JSON, delete from live DB
- Details: `no_agent` cron script using SessionDB API. Weekly (Sunday 06:00). Dry-run mode first
- Verification: Sessions inactive >30d archived, no data loss
- [ ] **CRON-03**: Daily report includes Jira ticket status via ngn-jira skill
- Details: Report queries Jira for tickets related to active sessions
- Verification: Daily report contains Jira ticket summaries
## Future (v1.2+)
- Archive restore script (JSON files are text-searchable; low urgency)
- On-demand repo cloning as dedicated skill
- Custom ngn-agent plugin package
## Out of Scope
- Custom scheduler — Hermes cron already handles this
- Custom memory provider implementation — Hindsight is production-ready and bundled
- Persistent Docker image with pre-cloned repos — would be large and stale quickly
## Traceability
| Requirement | Phase | Status |
|-------------|-------|--------|
| MEM-01 | TBD | Pending |
| REPO-01 | TBD | Pending |
| REPO-02 | TBD | Pending |
| SKIL-04 | TBD | Pending |
| CRON-01 | TBD | Pending |
| CRON-02 | TBD | Pending |
| CRON-03 | TBD | Pending |
---
*Requirements defined: 2026-06-14*
*Ready for roadmap: yes*