Skip to content
Agentmart
Submit a tool

atlas

Render a single "where am I, what have I done, what's next" survey HTML page composing anchor cascade + tesseract shelf + git state + CHECKPOINT.md + TaskList.

Skill by laurac8r

IndexedMIT★ 3
Write a review

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/laurac8r-atlas

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Render a single "where am I, what have I done, what's next" survey HTML page composing anchor cascade + tesseract shelf + git state + CHECKPOINT.md + TaskList. Read-only by design.

Use cases: design, frontend, research

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found
Runs as
Python package

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/atlas

What's inside

Skill name
atlas
Description
Render a single "where am I, what have I done, what's next" survey HTML page composing anchor cascade + tesseract shelf + git state + CHECKPOINT.md + TaskList. Read-only by design.
License
MIT
Instructions
190 lines in SKILL.md · view on GitHub

Before installing

  • Ships executable scripts

Files (7)

  • SKILL.md
  • __init__.py
  • baseline.html
  • rationalizations.md · evals/iteration-1/baseline/pressure/
  • agent-output.md · evals/iteration-1/baseline/quant/
  • rationalizations.md · evals/iteration-1/with-skill/pressure/
  • render.py

SKILL.md (first lines)

# /atlas

Survey skill — composes `/tesseract`'s anchor cascade + shelf reads,
`/checkpoint-resume`'s CHECKPOINT.md discovery, current git state, and the
agent's TaskList into one static HTML page at `~/.visual-aid/atlas-<slug>.html`.

**Read-only.** /atlas never writes to the tesseract shelf, bulk-beings,
CHECKPOINT.md, or the TaskList. The output HTML file is the only side effect.

## Arguments

- `--anchor <name>` — explicit anchor (passed as `override=` to
  `resolve_anchor`; skips the auto-cascade).
- `--all` — multi-anchor mode: drop the per-project sections (git, checkpoint,
  tasks); render the shelf only. v1 limitation; see ROADMAP.md Phase 5 follow-up
  entries for Option 2 (cwd-scoped) and Option 3 (per-anchor → project mapping).
- `--out <path>` — override the output path. Default is
  `~/.visual-aid/atlas-<anchor.slug>.html`.

## Process

Follow these steps in order. Collect warnings as you go; they populate
`AtlasInput.warnings` and surface in the rendered page — they never abort.

### 1. Resolve the anchor

```python
from pathlib import Path
from skills._shared.anchor import resolve_anchor
anchor, anchor_warns = resolve_anchor(Path.cwd(), override=args.anchor)
```

`resolve_anchor` walks a four-rung cascade (override → modified-file → branch →
memory), tags the source, slugifies, and returns `(Anchor, list[str])`. Always
call it. Do not hardcode an anchor slug — the cascade IS the contract;
reimplementing it loses the override semantics, the slugify rule, and the
warning population.

### 2. Read the shelf for that anchor

```python
from skills._shared.parse_shelf import parse_shelf
shelf_path = Path.home() / ".tesseract" / "shelf" / f"{anchor.slug}.md"
shelf_entries, shelf_warns = parse_shelf(shelf_path)
```

Missing file returns `([], [])`. Unreadable file returns `([], [warning])`.
Malformed entries are skipped with warnings. Append all returned warnings; keep
going.

### 3. Capture git state

```python
from skills._shared.parse_git import parse_git
git_state, git_warns = parse_git(Path.cwd())  # (None, []) when not a repo
```

Outside a repo, `git_state` is `None` — that is normal empty state, not a
failure.

### 4. Discover and parse CHECKPOINT.md

```python
from skills._shared.parse_checkpoint import parse_checkpoint
checkpoint, ckpt_warns = parse_checkpoint(Path.cwd())
```

Cascade: `cwd/CHECKPOINT.md` → `<repo-root>/CHECKPOINT.md` →
`<repo-root>/.worktrees/<branch-slug>/CHECKPOINT.md`. No file → `(None, [])`.
File found but unparseable → `CheckpointDoc(raw=text, branch="", ...)` plus a
warning — the renderer will display the raw text as a fallback.

### 5. Pull the TaskList (read-only)

Call the harness `TaskList()` tool. Map each entry to a `TaskRecord`:

```python
from skills.atlas.render import TaskRecord
raw_tasks = TaskList()
tasks = [
    TaskRecord(
        id=t["id"],
        subject=t["subject"],
        status=t["status"],
        blocked_by=t.get("blockedBy", []),
    )
    for t in raw_tasks
]
```

Do not call `TaskCreate`, `TaskUpdate`, or any other mutator during this step.
Read only.

**Fallback — Task tools absent from the tool surface.** Some sessions gate
`TaskCreate`/`TaskGet`/`TaskList`/`TaskUpdate` behind an experiment flag (legacy
`TodoWrite` can be absent too, and headless `-p` sessions are unaffected); when
that happens all four are simply missing, not erroring. Do not abort — fall back
to tracking tasks with a **Markdown table** instead:

| ID  | Subject | Status | Blocked by |
| --- | ------- | ------ | ---------- |

Build `TaskRecord`s from whatever Markdown to-do list is present in cwd (per
this repo's `CLAUDE.md › Tasks, Planning, and Execution`) and continue to step 6
as if `TaskList()` had returned them.

### 6. Build `AtlasInput` and call `render()`

```python
from skills.atlas.render import AtlasInput, render
inp = AtlasInput(
    anchor=anchor,
    shelf_entries=shelf_entries,
    git_state=git_state,
    checkpoint=checkpoint,
    task_list=tasks,
    mode="all" if args.all else "single",
    warnings=anchor_warns + shelf_warns + git_warns + ckpt_warns,
)

From laurac8r/claude-damn, MIT.

More from laurac8r/claude-damn

Signals

Stars
3
Forks
1
Open issues
0
Last commit (this folder)
Aug 5, 2026 (2 months ago)
Last release
Jul 19, 2026 (3 months ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from laurac8r

Cite this page

atlas skill. Agent Mart. https://agentmart.dev/skills/laurac8r-atlas (last updated 2026-10-24)