tzachbon/srulik-toolkit / plugins/srulik-toolkit/skills/to-project
to-project
VerifiedUse when the user wants to turn an idea, a notes dump, or a session transcript into a complete agent-ready local project, or to fold new input into a project that already exists. Triggers include "turn this into a project", "scaffold a project from this", "make a project out of this session/these notes", "set up a new repo for this", "extend an existing project with this", "add this to my project", "continue the project from these notes".
| File | Kind | Size |
|---|---|---|
| ADR-FORMAT.md | markdown | 2.7 KB |
| EXTEND.md | markdown | 4.0 KB |
| GLOSSARY-FORMAT.md | markdown | 2.8 KB |
| SKILL.md | skill md | 8.9 KB |
| templates.md | markdown | 6.8 KB |
templates.md
View on GitHub# Scaffold templates
File bodies the skill writes in step 6. Replace `{{placeholders}}`. Keep each
file lean: `AGENTS.md` is a map, not a manual.
---
## AGENTS.md
```markdown
# AGENTS.md
Guidance for any agent working in this repo.
## What this is
{{one-paragraph: what the project is, who it's for, what "done" looks like.}}
**Read `TASKS.md` first.** It is the live task board and the source of truth for
what is in flight.
## Working norms
{{project-specific norms inferred during the grill. Examples: build/test commands,
where output goes, review gates.}}
User-level and harness-level instructions still apply when present. Put only
project-specific conventions here so a fresh agent can work without access to
the scaffold author's personal configuration.
Everything stays tracked. Commit your work as you go; never leave the tree dirty
or important state uncommitted. `tidy-project` enforces this.
## You are not alone (multi-agent contract)
Assume other agents and the user work this repo at the same time.
- One **main/orchestrator session** runs the show. **Named side-agents** each take
a focused sub-task in their own context, often on their own branch or git
worktree.
- **`TASKS.md` is the coordination surface.** Claim a task before working it; mark
status as you go.
- **Shared files change under you.** Re-read right before editing, keep edits
narrow, never revert another writer's work.
- If a write is blocked or risks collision, work in a **worktree and merge back**
(rebase onto moved main, then `git merge --ff-only`).
- Converging a side session back into the main thread closes the loop; the main
session stays the source of truth.
## Skills (bundled)
Two skills live in `.agents/skills/`; `.claude/skills` is a relative alias to
that directory. Use them in normal work:
- **`capture-to-project`**: when a decision, term, durable fact, or actionable
item surfaces, run it to file the point into the right doc (`GLOSSARY.md`, an
ADR, `TASKS.md`, `docs/memory/`, or a `research/` note). It commits.
- **`tidy-project`**: occasionally (session start, or after a batch of changes),
run it to clear junk, fix stale tasks and index drift, then commit. It auto-fixes
safe items and proposes anything destructive.
## Index: where things live
`AGENTS.md` is the canonical steering file. `CLAUDE.md` imports it for Claude
Code.
- **`AGENTS.md`** (this file): steering instructions and directory map.
- **`TASKS.md`**: task board. Read first, update as you work.
- **`.agents/skills/`**: bundled project skills (capture/tidy).
`.claude/skills` is a relative alias for Claude Code; neither path points to a
global installation.
- **`assets/`**: all provided raw inputs (transcripts, images, scripts, the
originating dump). Tracked in git; provenance lives here.
- `assets/sources.md`: index of every provided asset.
- `assets/origin.md`: the originating idea (only if input was a typed prompt
with no file).
- **`research/`**: working and generated research the agents produce. Not for
provided source files (those go in `assets/`).
- **`docs/`**: durable docs.
- `docs/adr/`: architecture decision records (why a hard-to-reverse choice was
made). _(Created on the first hard-to-reverse trade-off. May not exist yet
in a fresh scaffold.)_
- `docs/memory/MEMORY.md`: repo-local memory index. One line per durable fact;
facts live in sibling files.
- **`GLOSSARY.md`**: glossary of the project's canonical terms. Challenge any
usage that conflicts with it. _(Created on the first term resolved during the
grill. May not exist yet in a fresh scaffold.)_
{{type-specific entries, e.g. for a presentation: "slides/: the deck".}}
{{If related projects were identified: "## Related projects\n- path/to/sibling: one-line why"}}
```
---
## TASKS.md
```markdown
# {{Project}}: Task Board
`TASKS.md` is the source of truth for what is in flight. Update it as you work.
## Now
_(empty: set the current focus when work starts)_
## Next
_(empty: queue the next items when priorities are clear)_
## Backlog
{{extracted goals as discrete, actionable items}}
## Open questions
{{questions surfaced during the grill that are not yet resolved}}
## Reference
- Origin and sources: `assets/sources.md`
- Glossary: `GLOSSARY.md` _(if present; may not exist yet in a fresh scaffold)_
- Decisions: `docs/adr/` _(if present; may not exist yet in a fresh scaffold)_
```
---
## assets/sources.md
```markdown
# Sources: intake index
One row per source fed into this project. Newest first.
| Date | Source | Path | Notes |
|---|---|---|---|
| {{YYYY-MM-DD}} | {{origin: session / notes / artifact}} | {{assets/<file>}} | {{one line}} |
```
---
## assets/origin.md (typed-only idea, no artifact)
```markdown
# Origin
Captured {{YYYY-MM-DD}}. The idea this project was scaffolded from, verbatim.
{{the user's original idea text}}
```
---
## docs/memory/MEMORY.md
```markdown
# Memory Index
Repo-local durable facts. One line per fact; the fact itself lives in a sibling
file in this directory. Format for each file:
---
name: <kebab-slug>
description: <one-line summary used to judge relevance>
---
<the fact. Link related facts with [[their-name]].>
Add a pointer here when you write a fact: `- [Title](file.md): hook`.
<!-- no facts yet -->
```
---
## .gitignore by project type
These blocks are examples. Include the common block always, then append the
block closest to the declared type. For free-form types with no exact match,
compose from the common block plus the nearest example.
Always include the common block. Append the type-specific block.
### common
```
.DS_Store
*.log
.env
.env.local
.scratch/
tmp/
```
### presentation (Slidev / node)
```
node_modules/
dist/
.slidev/
*.local
```
### software (node)
```
node_modules/
dist/
build/
coverage/
```
### software (python)
```
__pycache__/
*.pyc
.venv/
venv/
.pytest_cache/
dist/
build/
*.egg-info/
```
### research / docs / ops-runbook
```
# (common block is usually enough; add tool-specific dirs as they appear)
```
---
## Type-specific extensions
The fixed core (AGENTS.md, TASKS.md, assets/, docs/, docs/memory/, research/) is
always created. On top of it, add the extension for the declared type. Keep it
minimal: an empty dir with a `.gitkeep`, not a code skeleton. Omit when unsure and
let structure emerge. For a free-form type with no row, compose from the nearest
match or skip.
| Type | Add | AGENTS.md index entry |
|---|---|---|
| software | `src/`, `tests/` | "src/: source. tests/: tests." |
| presentation | `slides/` | "slides/: the deck." |
| research | (none beyond core) | research/ already covers it |
| docs | (none beyond core) | docs/ already covers it |
| ops-runbook | `runbooks/` | "runbooks/: the procedures." |
| library | `src/`, `tests/`, `examples/` | "src/, tests/, examples/." |
| data-analysis | `data/`, `notebooks/` | "data/: inputs. notebooks/: analysis." |
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |