@zalom/plastic 1.0.0-alpha.2 → 1.0.0-alpha.20

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
Files changed (75) hide show
  1. package/PLASTIC.md +128 -473
  2. package/README.md +90 -58
  3. package/agents/future-intent-researcher.md +1 -1
  4. package/agents/intent-curator.md +1 -1
  5. package/bin/plastic.js +57 -0
  6. package/bin/test +28 -0
  7. package/deprecations.yml +7 -6
  8. package/hooks/auto-arm +5 -0
  9. package/hooks/bash-gate +3 -0
  10. package/hooks/check-update +12 -8
  11. package/hooks/code-gate +10 -0
  12. package/hooks/hooks.json +25 -4
  13. package/hooks/statusline +50 -10
  14. package/package.json +2 -2
  15. package/scripts/dashboard.rb +480 -0
  16. package/scripts/doctor.rb +950 -0
  17. package/scripts/hook-auto-arm +52 -0
  18. package/scripts/hook-bash-gate +53 -0
  19. package/scripts/hook-code-gate +39 -0
  20. package/scripts/hook-continue +15 -114
  21. package/scripts/hook-gate-check +19 -4
  22. package/scripts/hook-session-start +76 -31
  23. package/scripts/install.rb +91 -480
  24. package/scripts/lib/bridge.rb +255 -0
  25. package/scripts/lib/installer_core.rb +760 -0
  26. package/scripts/migrate-to-global +1 -1
  27. package/scripts/select-update-target +93 -0
  28. package/scripts/uninstall.rb +53 -0
  29. package/scripts/update.rb +142 -0
  30. package/scripts/versions.rb +141 -0
  31. package/skills/_active-intent-gate.md +26 -0
  32. package/skills/auto/SKILL.md +62 -9
  33. package/skills/auto/evals/evals.json +92 -0
  34. package/skills/auto/references/agent-architecture.md +60 -0
  35. package/skills/brainstorming/SKILL.md +143 -0
  36. package/skills/brainstorming-grill-me/SKILL.md +5 -5
  37. package/skills/continuing/SKILL.md +20 -4
  38. package/skills/continuing/references/context-management.md +32 -0
  39. package/skills/creating-intent/SKILL.md +16 -1
  40. package/skills/creating-intent/references/lifecycle.md +74 -0
  41. package/skills/creating-intent/references/wikilinks.md +8 -0
  42. package/skills/creating-project/SKILL.md +8 -4
  43. package/skills/creating-project/references/hubs-projects.md +55 -0
  44. package/skills/dashboard/SKILL.md +92 -0
  45. package/skills/doctor/SKILL.md +116 -0
  46. package/skills/doctor/references/gates-stuck-detection.md +38 -0
  47. package/skills/doctor/report.md +96 -0
  48. package/skills/evaluating-skills/SKILL.md +140 -0
  49. package/skills/evaluating-skills/assets/eval-template.json +12 -0
  50. package/skills/evaluating-skills/evals/evals.json +75 -0
  51. package/skills/evaluating-skills/references/convention-checks.md +76 -0
  52. package/skills/evaluating-skills/references/eval-methodology.md +154 -0
  53. package/skills/executing-plan/SKILL.md +3 -3
  54. package/skills/install/SKILL.md +56 -8
  55. package/skills/intent-curator/SKILL.md +3 -3
  56. package/skills/linking-intents/SKILL.md +5 -1
  57. package/skills/linking-intents/references/zettelkasten.md +33 -0
  58. package/skills/managing-index/SKILL.md +5 -1
  59. package/skills/releasing/SKILL.md +119 -18
  60. package/skills/releasing/references/deprecations.md +44 -0
  61. package/skills/research/SKILL.md +114 -0
  62. package/skills/savepoint/SKILL.md +46 -37
  63. package/skills/savepoint/references/context-management.md +32 -0
  64. package/skills/uninstall/SKILL.md +39 -28
  65. package/skills/update/SKILL.md +41 -36
  66. package/skills/versions/SKILL.md +65 -0
  67. package/skills/writing-instructions/SKILL.md +159 -0
  68. package/skills/writing-instructions/references/agentskills-spec.md +135 -0
  69. package/skills/writing-plans/SKILL.md +183 -0
  70. package/templates/agents.md +16 -0
  71. package/templates/outcome.md +13 -0
  72. package/templates/project.yml +5 -0
  73. package/templates/savepoint.md +14 -13
  74. package/templates/spec.md +25 -0
  75. package/bin/install.js +0 -29
@@ -0,0 +1,114 @@
1
+ ---
2
+ name: plastic-research
3
+ description: "Research a topic for the active intent. Agent decides shallow vs deep based on scope. Produces reports in the intent's resources/ directory."
4
+ ---
5
+
6
+ # Research
7
+
8
+ Investigate a topic related to the active intent. Choose the right depth, produce a cited report.
9
+
10
+ **Announce at start:** "I'm researching [topic] for intent {id} — {name}."
11
+
12
+ ## Active Intent Gate
13
+
14
+ Before proceeding, resolve the active intent:
15
+
16
+ 1. **Detect store:** Read `~/.plastic/projects.yml`, match CWD against registered project paths. If match → project store at `~/.plastic/projects/{slug}/store/`. If no match → global store at `~/.plastic/store/`.
17
+ 2. **Find active intent:** Read `INDEX.md` from the detected store. Look under `## Active`. If exactly one → use it. If multiple → ask which. If none → refuse: "No active intent. Create one first with /plastic-creating-intent"
18
+ 3. **Resolve intent directory:** `{store}/store/{id}--{slug}/`
19
+
20
+ ## Depth Decision
21
+
22
+ Before starting research, assess the question against these criteria:
23
+
24
+ ### Shallow Research
25
+ Use when:
26
+ - The question is narrow and factual (e.g., "what's the API for X?", "does library Y support Z?")
27
+ - A single source or quick search is sufficient
28
+ - The answer is likely well-documented
29
+ - Time budget: minutes, not hours
30
+
31
+ **Method:** Direct web search + codebase exploration + single-pass synthesis.
32
+
33
+ ### Deep Research
34
+ Use when:
35
+ - The question needs multiple sources and cross-verification
36
+ - It's a landscape survey, competitive analysis, or architectural decision
37
+ - The user explicitly asks for thorough/exhaustive/comprehensive research
38
+ - Conflicting information exists and needs adversarial verification
39
+ - Time budget: significant, user expects depth
40
+
41
+ **Method:** Delegate to the native agent's deep research capability.
42
+ - On Claude Code: invoke the `deep-research` skill via `Skill` tool
43
+ - On other agents: use the agent's equivalent (see intent 7 for cross-agent mapping)
44
+ - If no native deep research exists: fan out multiple web searches manually, cross-reference, synthesize
45
+
46
+ ### Decision Heuristic
47
+
48
+ Ask yourself:
49
+ 1. Can I answer this with 1-2 searches? → **Shallow**
50
+ 2. Do I need to compare 3+ sources? → **Deep**
51
+ 3. Is the user asking for a survey or landscape view? → **Deep**
52
+ 4. Am I fact-checking a specific claim? → **Shallow**
53
+ 5. Could wrong information here cause architectural mistakes? → **Deep**
54
+
55
+ Announce your choice: "This needs [shallow/deep] research because [reason]."
56
+
57
+ ## Output
58
+
59
+ ### File Location
60
+ `{intent_dir}/resources/{type}--{topic}.md`
61
+
62
+ Create the resources/ directory if it doesn't exist.
63
+
64
+ ### Naming Convention
65
+ - `{type}` — one of: `deep-research`, `competitive-analysis`, `technical-spike`, `reference`, `landscape-survey`
66
+ - `{topic}` — kebab-case description of the research subject
67
+
68
+ Examples:
69
+ - `deep-research--installer-patterns.md`
70
+ - `technical-spike--sqlite-vec-performance.md`
71
+ - `reference--claude-code-hooks-api.md`
72
+ - `landscape-survey--ai-agent-ecosystem.md`
73
+
74
+ ### Report Structure
75
+
76
+ ```markdown
77
+ # {Type}: {Topic}
78
+
79
+ **Intent:** {id} — {name}
80
+ **Depth:** shallow | deep
81
+ **Date:** YYYY-MM-DD
82
+
83
+ ## Summary
84
+ [2-3 sentence executive summary]
85
+
86
+ ## Findings
87
+
88
+ ### [Finding 1 title]
89
+ [Details with inline citations where applicable]
90
+
91
+ ### [Finding 2 title]
92
+ ...
93
+
94
+ ## Sources
95
+ - [Source 1 title](url) — what was learned from this source
96
+ - [Source 2 title](url) — ...
97
+
98
+ ## Relevance to Intent
99
+ [How these findings affect the active intent's decisions or design]
100
+ ```
101
+
102
+ ## After Research
103
+
104
+ 1. Commit to store repo:
105
+ ```
106
+ cd {store_root} && git add . && git commit -m "research: {type} — {topic} (intent {id})"
107
+ ```
108
+
109
+ 2. Log in intent's `## Insights`:
110
+ ```
111
+ - Research: {type} — {topic}. Key finding: [one sentence]. See resources/{filename}.
112
+ ```
113
+
114
+ 3. Research does NOT chain to any other skill. It's a side quest — the user decides what to do with the findings.
@@ -1,57 +1,66 @@
1
1
  ---
2
- name: plastic:savepoint
3
- description: Use when context is being compacted (PreCompact hook), user says "save" or "savepoint", or before ending a session. Saves the active intent's state so work can resume after /clear.
2
+ name: plastic-savepoint
3
+ description: Use when verifying or repairing an intent's savepoint ledger, when the user says "save" or "savepoint", or when a PreCompact hook fires. The ledger is written automatically by the gate hook at each lifecycle boundary; this skill only reads, verifies, and rebuilds it.
4
4
  ---
5
5
 
6
6
  # Savepoint
7
7
 
8
- ## When to Use
9
- - PreCompact hook fires (automatic)
10
- - User says "save", "savepoint", or "save progress"
11
- - Before ending a long session
12
- - Before switching to a different intent
8
+ `savepoint.md` is a deterministic, append-only ledger of an intent's cycle steps, one line
9
+ per lifecycle milestone, newest at the bottom:
13
10
 
14
- ## Workflow
11
+ ```
12
+ 2026-06-16T14:02Z What 34--cycle-step-savepoints.md
13
+ 2026-06-16T14:20Z Why spec.md created
14
+ 2026-06-16T15:10Z How plan.md created
15
+ 2026-06-16T15:11Z How checklist.md created
16
+ 2026-06-16T16:40Z Exec outcome.md created
17
+ ```
15
18
 
16
- ### 1. Find Active Intent(s)
17
- Read `~/.plastic/INDEX.md` (or the project INDEX.md) and extract all intents listed under `## Active`.
19
+ It is **sugar on top of the conventions**, not a source of truth. The gate hook
20
+ (`hook-gate-check`) writes each line automatically when a stage file is created, so there is
21
+ nothing to save by hand. State is always derivable from files-on-disk; the ledger just lets
22
+ a resuming agent read the cycle's succession from one glance (last line = where we are).
18
23
 
19
- ### 2. For Each Active Intent
20
- Read the intent directory at `~/.plastic/store/ID--slug/`:
24
+ ## When to Use
25
+ - A PreCompact hook fires, or the user says "save" / "savepoint": verify the ledger is current.
26
+ - Resuming an intent: read the ledger to learn the cycle's succession quickly.
27
+ - The ledger looks stale, empty, or missing: rebuild it from the filesystem.
21
28
 
22
- **a. Update checklist.md** (if exists):
23
- - Check off completed items
24
- - Add any new items discovered during the session
29
+ ## What this skill does NOT do
30
+ - It does not write prose, "in progress", "next step", or "blockers". The old 50%-context
31
+ prose savepoint is retired. The semantic trace lives in `## Insights` (append-only,
32
+ newest at the bottom); the byte history lives in git.
25
33
 
26
- **b. Create/update savepoint.md:**
27
- ```markdown
28
- # Savepoint
34
+ ## Workflow
29
35
 
30
- ## Last Updated
31
- {{DATE}} — Session #{{N}}
36
+ ### 1. Find Active Intent(s)
37
+ Read the active store's `INDEX.md` and extract intents under `## Active`.
32
38
 
33
- ## In Progress
34
- - (what was being worked on when savepoint triggered)
35
- - Next: (immediate next step)
39
+ ### 2. For Each Active Intent — verify the ledger
40
+ - Read `savepoint.md`. Confirm it exists and is non-empty.
41
+ - Confirm the **last line's stage** matches the stage derived from files-on-disk
42
+ (`Bridge.derive_stage`). If they disagree, or the file is missing/empty, the ledger has
43
+ drifted.
36
44
 
37
- ## Blockers
38
- (any blockers or open questions)
45
+ ### 3. Rebuild on drift
46
+ Reconstruct from the filesystem rather than hand-editing:
39
47
 
40
- ## Key Discoveries This Session
41
- - (important things learned)
48
+ ```bash
49
+ ruby -r ~/.plastic/scripts/lib/bridge -e \
50
+ 'Bridge.rebuild_savepoint("<intent_dir>")'
42
51
  ```
43
52
 
44
- **c. Update `{ID}--{slug}.md`:**
45
- - Add observations to `## Insights` section
46
-
47
- ### 3. Update INDEX.md
48
- Verify the `## Active` section is accurate.
53
+ This rewrites `savepoint.md` from the milestone files present (timestamps from mtimes),
54
+ in stage order. Safe to run anytime: the ledger is derived.
49
55
 
50
- ### 4. Commit
56
+ ### 4. Commit (store only)
57
+ If a rebuild changed the ledger, commit it in the store repo:
51
58
  ```bash
52
- git add .plastic/
53
- git commit -m "chore: savepoint — [active intent name]"
59
+ cd <store-root> && git add . && git commit -m "chore: rebuild savepoint ledger — [intent]"
54
60
  ```
61
+ Never push `~/.plastic/`.
62
+
63
+ ## References
55
64
 
56
- ### 5. Notify User
57
- Tell the user: "Context is getting large. I've saved progress to intent [ID] — [name]. Please run `/clear` and say `continue` to resume."
65
+ - Read `references/context-management.md` for the full save/continue protocol and how the
66
+ resume flow consumes the ledger.
@@ -0,0 +1,32 @@
1
+ # Context Management (Start-Save-Continue)
2
+
3
+ ## Save Point
4
+ Triggered by PreCompact hook or manually:
5
+ 1. Find active intent(s) from `~/.plastic/INDEX.md`
6
+ 2. Update active intent's `checklist.md` (check off completed items)
7
+ 3. Update active intent's `savepoint.md` (in-progress, next steps, blockers, discoveries)
8
+ 4. Add observations to `## Insights`
9
+ 5. Update INDEX.md
10
+ 6. Commit: `cd ~/.plastic && git add . && git commit -m "chore: savepoint — [intent name]"`
11
+ 7. Notify user to `/clear`
12
+
13
+ ## Continue
14
+ Triggered by UserPromptSubmit hook when user says "continue". Priority order:
15
+
16
+ **1. Active intents first (resume work):**
17
+ 1. Read INDEX.md → find active intent(s)
18
+ 2. Read active intent's `intent.md` → what and why
19
+ 3. Read active intent's `savepoint.md` → where we left off
20
+ 4. Read active intent's `checklist.md` → what's next
21
+ 5. Announce: intent name, current state, next step, blockers
22
+ 6. Resume
23
+
24
+ **2. No active intents → offer future intents:**
25
+ 1. List all future intents from INDEX.md
26
+ 2. Present them as options
27
+ 3. When user picks one, move to Active in INDEX.md
28
+
29
+ **3. Stale future intents (untouched 3+ days) → triage:**
30
+ - **activate** — start working on it now
31
+ - **abandon** — mark as abandoned
32
+ - **defer to agent** — implement, research, or ideate
@@ -1,48 +1,59 @@
1
1
  ---
2
- name: plastic:uninstall
3
- description: Use when the user wants to remove Plastic. Deregisters the plugin from the current agent (Claude Code, Cursor, etc.) and optionally deletes all data.
2
+ name: plastic-uninstall
3
+ description: Use when the user wants to remove Plastic from an agent. Runs the manifest-driven uninstaller (removes skills, hooks, statusline, and any legacy plugin layout), reports exactly what was removed and what was left behind, then gives verification steps. Optionally deletes the intent store.
4
4
  ---
5
5
 
6
6
  # Uninstall Plastic
7
7
 
8
- ## Procedure
8
+ The installer tracks every file it writes in a manifest, so uninstall is exact and
9
+ leaves no orphans. Prefer running it through the CLI; this skill wraps the same
10
+ underlying uninstaller and adds reporting + verification.
9
11
 
10
- ### Step 1: Detect the current agent
12
+ ## Procedure
11
13
 
12
- Check environment variables to determine which agent is running:
13
- - `$CLAUDE_PLUGIN_ROOT` → Claude Code
14
- - `$CURSOR_PLUGIN_ROOT` → Cursor
15
- - Otherwise → unknown agent
14
+ ### Step 1: Run the uninstaller
16
15
 
17
- ### Step 2: Deregister from the agent
16
+ ```bash
17
+ npx @zalom/plastic@latest uninstall --claude
18
+ ```
18
19
 
19
- **Claude Code:**
20
- - Remove `"plastic"` from `extraKnownMarketplaces` in `~/.claude/settings.json`
21
- - Remove `"plastic@plastic"` from `enabledPlugins` in `~/.claude/settings.json`
22
- - Announce: "Plastic deregistered from Claude Code."
20
+ (Use `--codex` / `--hermes` / `--all` to target other agents. `bunx` works too.)
23
21
 
24
- **Other agents:** Provide manual instructions for their deregistration process.
22
+ This removes, for the targeted agent:
23
+ - all `~/.claude/skills/plastic-*/` skills
24
+ - all `~/.claude/hooks/plastic-*` hooks (and restores any saved original statusline)
25
+ - the `~/.claude/plastic/` state dir + manifest
26
+ - any **legacy** plugin layout: `plugins/marketplaces/plastic`, `plugins/cache/plastic`,
27
+ the old nested `skills/plastic/`, and the `plastic@plastic` /
28
+ `extraKnownMarketplaces.plastic` / `known_marketplaces.json` registrations
25
29
 
26
- ### Step 3: Offer the data decision
30
+ ### Step 2: Report removed vs left
27
31
 
28
- Present clearly:
32
+ Relay the uninstaller's output to the user — what was **removed** and what was
33
+ **left in place**:
34
+ - **Left:** `~/.plastic/` (intent store, history, projects) and any non-Plastic
35
+ settings.json entries.
29
36
 
30
- ```
31
- Plastic is deregistered from [agent name].
37
+ ### Step 3: Verify removal
32
38
 
33
- Your intent store at ~/.plastic/ is untouched.
39
+ Tell the user to confirm:
34
40
 
35
- Would you like to delete your Plastic data now?
41
+ ```bash
42
+ ls ~/.claude/skills | grep '^plastic-' # → no output
43
+ ls ~/.claude/hooks | grep '^plastic-' # → no output
44
+ grep -n plastic ~/.claude/settings.json # → no plastic hook/plugin refs
45
+ ```
36
46
 
37
- ⚠️ WARNING: This permanently removes ALL intents, history,
38
- and any projects in ~/.plastic/projects/.
39
- This is irreversible.
47
+ ### Step 4: Offer the data decision
40
48
 
41
- a) Keep everything (recommended) — you can re-install Plastic later
42
- b) Delete everything now — removes ~/.plastic/ entirely
43
49
  ```
50
+ Plastic is uninstalled from [agent].
51
+ Your intent store at ~/.plastic/ is untouched.
44
52
 
45
- ### Step 4: Execute user's choice
53
+ Delete it too?
54
+ a) Keep everything (recommended) — re-install anytime with npx
55
+ b) Delete everything now — removes ~/.plastic/ entirely (irreversible)
56
+ ```
46
57
 
47
- - **Keep:** Done. Tell the user: "Your data is at ~/.plastic/. Re-install anytime with `/plastic:install`."
48
- - **Delete:** Run `rm -rf ~/.plastic/` and confirm: "Plastic data deleted."
58
+ - **Keep:** "Your data is at ~/.plastic/. Re-install anytime with `npx @zalom/plastic@latest install --claude`."
59
+ - **Delete:** run `rm -rf ~/.plastic/` and confirm.
@@ -1,6 +1,6 @@
1
1
  ---
2
- name: update
3
- description: Use when updating Plastic after a plugin update, or when the user says "update plastic". Runs the npx installer to sync core files and re-register agent adapters.
2
+ name: plastic-update
3
+ description: Use when updating Plastic. Runs the `update` verb, which reads the installed VERSION, derives its channel, queries npm dist-tags, and advances to the next version on that channel (or switches channel with a flag).
4
4
  ---
5
5
 
6
6
  # Update Plastic
@@ -8,62 +8,67 @@ description: Use when updating Plastic after a plugin update, or when the user s
8
8
  ## When to Use
9
9
  - User says "update plastic", "sync plastic", or "upgrade plastic"
10
10
  - Statusline shows "Plastic update available"
11
- - After a version bump notification
11
+ - After a version-bump notification
12
12
 
13
- ## Prerequisites
13
+ ## What it does
14
14
 
15
- Global install must exist (`~/.plastic/INDEX.md` present). If not, tell the
16
- user to run `npx @zalom/plastic@latest` first.
15
+ `update` is a single deterministic command. It reads `~/.plastic/VERSION`, derives the
16
+ channel from the version string (`-alpha`/`-beta`/none → stable), queries `npm` dist-tags,
17
+ and advances to the **next version on the current channel**. "Already up to date" is a
18
+ clean no-op. You do not compute the target yourself — the script does.
17
19
 
18
- ## Procedure
20
+ ## Flags
19
21
 
20
- ### Step 1: Verify global install exists
22
+ | Flag | Behaviour |
23
+ |------|-----------|
24
+ | (none) | Advance to the next version on the **current** channel |
25
+ | `--latest` | Switch to / advance the **stable** channel (toward stability — frictionless) |
26
+ | `--beta` | Switch to / advance the **beta** channel |
27
+ | `--alpha` | Switch to / advance the **alpha** channel (bleeding edge — confirmed if moving down in stability) |
21
28
 
22
- ```bash
23
- if [ ! -f ~/.plastic/INDEX.md ]; then
24
- echo "No global install found. Run: npx @zalom/plastic@latest"
25
- exit
26
- fi
27
- ```
29
+ Switching toward a more stable channel is frictionless; switching toward bleeding edge is
30
+ confirmed. To roll **back** to a previously-installed version, use `plastic-versions`.
31
+
32
+ ## Prerequisites
28
33
 
29
- ### Step 2: Run the installer
34
+ Plastic must be installed (`~/.plastic/VERSION` present). If not, run
35
+ `npx @zalom/plastic install --claude` first.
36
+
37
+ ## Procedure
38
+
39
+ ### Step 1: Run the update
30
40
 
31
41
  ```bash
32
- npx @zalom/plastic@latest --claude
42
+ npx @zalom/plastic update # next version on the current channel
43
+ # or: npx @zalom/plastic update --beta / --latest / --alpha
33
44
  ```
34
45
 
35
- This re-runs the installer which:
36
- - Downloads the latest version from npm
37
- - Syncs core files (PLASTIC.md, scripts, hooks) to ~/.plastic/
38
- - Re-registers hooks and skills into Claude Code's ~/.claude/
39
- - Preserves all user data (INDEX.md, config.yml, projects.yml, store/)
46
+ Append the agent flag(s) if not just Claude (`--codex`, `--hermes`, `--all`).
47
+ The command prints the transition (`vX → vY`) or "already up to date", and records the
48
+ move in the append-only `~/.plastic/versions.json` ledger.
40
49
 
41
- ### Step 3: Announce key changes
50
+ ### Step 2: Announce key changes
42
51
 
43
- After the installer completes, read `~/.plastic/PLASTIC.md` and announce any
44
- convention changes that affect the current session. This corrects the agent's
45
- in-context understanding without needing /clear.
52
+ After it completes, read `~/.plastic/PLASTIC.md` and announce convention changes that
53
+ affect the current session:
46
54
 
47
- Format:
48
55
  ```
49
- Plastic updated to vX.Y.Z.
56
+ Plastic updated to vX.Y.Z (channel).
50
57
 
51
- Key changes in this version:
52
- - [list notable convention changes if any]
58
+ Key changes:
59
+ - [notable convention changes, if any]
53
60
 
54
61
  Recommendation: run /clear for a clean session with all new conventions loaded.
55
62
  ```
56
63
 
57
- ### Step 4: Commit
64
+ ### Step 3: Health check
58
65
 
59
- ```bash
60
- cd ~/.plastic && git add PLASTIC.md scripts/ AGENTS.md VERSION 2>/dev/null && git commit -m "chore: update Plastic core files" --allow-empty
61
- ```
66
+ Invoke `plastic-doctor`. If all checks pass: **"Health check: all clear."** Otherwise show
67
+ the report and offer to fix.
62
68
 
63
- ### Step 5: Clear update cache
69
+ ### Step 4: Commit + clear update cache
64
70
 
65
71
  ```bash
72
+ cd ~/.plastic && git add PLASTIC.md scripts/ AGENTS.md VERSION versions.json 2>/dev/null && git commit -m "chore: update Plastic to $(cat ~/.plastic/VERSION)" --allow-empty
66
73
  rm -f ~/.plastic/.cache/update-check.json
67
74
  ```
68
-
69
- This removes the statusline warning since the update is now applied.
@@ -0,0 +1,65 @@
1
+ ---
2
+ name: plastic-versions
3
+ description: Use when the user wants to see their Plastic version history or roll back to a previously-installed version after a bad release. Manages the local, append-only versions.json ledger and steps between versions the user has actually run. For moving to a brand-new release, use plastic-update instead.
4
+ ---
5
+
6
+ # Plastic Versions — local version time-machine
7
+
8
+ ## When to Use
9
+ - "show plastic versions", "version history", "what versions have I run"
10
+ - "roll back plastic", "downgrade", "go back to the version that worked", "revert plastic"
11
+ - A new version broke something and the user wants their last known-good build
12
+
13
+ For upgrading to a **new** release, use `plastic-update`. This skill only navigates
14
+ versions you have **already installed** — the ones recorded in the ledger.
15
+
16
+ ## The ledger
17
+
18
+ `~/.plastic/versions.json` is an **append-only JSONL** ledger — one line per version
19
+ change, never modified or deleted:
20
+
21
+ ```json
22
+ {"version":"1.0.0-alpha.17","action":"install","at":"..."}
23
+ {"version":"1.0.0-alpha.18","action":"update","at":"..."}
24
+ ```
25
+
26
+ `action` ∈ `install | reinstall | update | downgrade` (derived from version direction; the
27
+ channel is derived from the version string, never stored). It is a troubleshooting record.
28
+
29
+ ## Procedure
30
+
31
+ ### Show history
32
+
33
+ ```bash
34
+ npx @zalom/plastic versions
35
+ ```
36
+
37
+ Prints the table with the currently-installed version marked. If the most recent action was
38
+ a `downgrade`, it asks whether to keep rolling back.
39
+
40
+ ### Roll back
41
+
42
+ ```bash
43
+ npx @zalom/plastic versions --downgrade # one step back
44
+ npx @zalom/plastic versions --downgrade --version 1.0.0-alpha.15 # to a specific run
45
+ ```
46
+
47
+ Rollback targets are restricted to versions in the ledger (only builds you have actually
48
+ run, so you only ever return to something known-good for you). The chosen version is
49
+ re-fetched from npm and re-synced; your intent store, config, and the ledger are untouched.
50
+
51
+ ### Step forward (after a rollback)
52
+
53
+ ```bash
54
+ npx @zalom/plastic versions --upgrade # one step forward in your history
55
+ ```
56
+
57
+ ### After any change
58
+
59
+ Run `plastic-doctor` to confirm health, and suggest `/clear` so the session picks up the
60
+ swapped conventions.
61
+
62
+ ## Notes
63
+ - The ledger is **never** edited or pruned — it is the audit trail.
64
+ - Downgrades cannot un-migrate a store-format change; if a warning appears, surface it to
65
+ the user rather than forcing the rollback.