From fpl-hsmem
Initialize Hindsight memory for a new project. Sets the bank mission, ingests existing formal artifacts (PRDs, RFCs, ADRs), and creates a small starter set of mental models. Use when the user says "set up memory for this project", "bootstrap hindsight here", "initialize Hindsight for <repo>", or when memory_status shows an empty bank in an active project.
How this skill is triggered — by the user, by Claude, or both
Slash command
/fpl-hsmem:bootstrapThis skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
One-time setup for a Hindsight bank when a project is newly activated.
One-time setup for a Hindsight bank when a project is newly activated.
This is not the technical wiring (.mcp.json, hooks — that's done by
setup.js); this is the memory seeding layer.
Skip and report if any of these are true:
memory_status returns "unreachable" — tell the user to start Hindsight firstmental_model_list already shows pages — the bank is initialized, only
proceed if the user confirms re-bootstrapcwd is not a real project root — ask the user to confirmmemory_get_current_bank → confirm the bank ID with the userREADME.md / project's CLAUDE.md if present to understand
the project's purposememory_set_mission with the one-sentence description.
Example: "ForgePlan workflow tooling — focus on memory, plugin, and
marketplace decisions."Look for these locations and ingest each found file:
forge/prds/*.mdforge/rfcs/*.mdforge/adrs/*.md or docs/decisions/*.mddocs/architecture.md (top-level only)For each: document_ingest_file(path).
Skip: archived/old folders, generated docs, vendor folders.
Propose to the user 2-3 mental models that fit this project. Examples:
decisions-log — "What architectural / product decisions have we made and why?"tech-debt — "What technical debt have we identified but not yet addressed?"team-conventions — "What conventions / patterns are specific to this codebase?"incident-history — "What incidents have we hit and how were they resolved?" (for ops-heavy projects)Only create what the user confirms. Less is more — empty mental models create noise; full ones come from real conversation history.
Print a one-line summary:
Bootstrap complete:
- Mission set
- N documents ingested
- M mental models created
- Bank ready for auto-recall / auto-retain
Read / Grep for code)..env files, or credentials.npx claudepluginhub forgeplan/marketplace --plugin fpl-hsmemGuides creation, editing, and verification of skills for AI coding agents using test-driven development with subagent scenarios. Use when authoring or debugging skills.