martin-loop 0.1.3 → 0.1.4

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.
@@ -1,126 +1,126 @@
1
- # Examples
2
-
3
- These examples are grounded in the current CLI and MCP surfaces in this repo. Where an example depends on a real provider path, it is labeled that way explicitly.
4
-
5
- These are still primarily repo-local RC examples. The root `martin-loop` package facade is now real and smoke-validated, but registry publication remains a later release step.
6
-
7
- ## 1. Stub-backed hello world
8
-
9
- Use this when you want a safe first pass through the loop without real model spend.
10
-
11
- ### PowerShell
12
-
13
- ```powershell
14
- $env:MARTIN_LIVE='false'
15
- pnpm run:cli -- run `
16
- --workspace ws_demo `
17
- --project proj_demo `
18
- --objective "Describe the current Martin run lifecycle in one paragraph" `
19
- --verify "pnpm --filter @martin/core test"
20
- Remove-Item Env:MARTIN_LIVE
21
- ```
22
-
23
- Why this is useful:
24
-
25
- - exercises `runMartin`
26
- - writes a real loop record and artifacts
27
- - avoids external provider dependencies
28
-
29
- ## 2. Repo-backed task with explicit scope
30
-
31
- Use allow and deny paths so the task contract is narrow and reviewable.
32
-
33
- ```bash
34
- pnpm run:cli -- run \
35
- --cwd . \
36
- --objective "Tighten README wording for the OSS quickstart" \
37
- --verify "pnpm --filter @martin/core test" \
38
- --allow-path README.md \
39
- --allow-path docs/oss/** \
40
- --deny-path apps/control-plane/** \
41
- --accept "Only update documentation files" \
42
- --accept "Do not modify runtime code"
43
- ```
44
-
45
- What this demonstrates:
46
-
47
- - repo root selection with `--cwd`
48
- - scoped file-edit boundaries
49
- - acceptance criteria injection into the task contract
50
-
51
- ## 3. Safety-block example
52
-
53
- This example is expected to block before execution because the verifier command is unsafe.
54
-
55
- ```bash
56
- pnpm run:cli -- run \
57
- --objective "Try to run an unsafe verifier" \
58
- --verify "rm -rf ."
59
- ```
60
-
61
- Expected behavior:
62
-
63
- - the leash blocks the verifier command before adapter execution
64
- - the run exits through a safety-oriented path rather than pretending the command was acceptable
65
- - the attempt artifact set includes a persisted leash artifact when applicable
66
-
67
- The point of this example is not that `rm` exists on every machine. The point is that the raw verifier text is evaluated before the process would be allowed to run.
68
-
69
- ## 4. Budget-constrained live run
70
-
71
- This is a live-provider example. Only use it when you have the relevant CLI and credentials configured.
72
-
73
- ```bash
74
- pnpm run:cli -- run \
75
- --engine codex \
76
- --model o3 \
77
- --objective "Refactor the CLI argument parser for clarity" \
78
- --verify "pnpm --filter @martin/cli test" \
79
- --budget-usd 2 \
80
- --soft-limit-usd 1 \
81
- --max-iterations 2
82
- ```
83
-
84
- What to review afterward:
85
-
86
- - admission and settlement events in `ledger.jsonl`
87
- - cost provenance labels in the run artifacts
88
- - whether the loop stopped for completion, budget pressure, or lack of progress
89
-
90
- ## 5. MCP invocation shape
91
-
92
- The MCP server exposes `martin_run`, `martin_inspect`, and `martin_status`.
93
-
94
- Example `martin_run` payload:
95
-
96
- ```json
97
- {
98
- "objective": "Tighten the local dashboard copy",
99
- "workingDirectory": ".",
100
- "engine": "claude",
101
- "verificationPlan": ["pnpm --filter @martin/control-plane test"],
102
- "maxUsd": 5,
103
- "maxIterations": 2,
104
- "maxTokens": 20000,
105
- "workspaceId": "ws_mcp",
106
- "projectId": "proj_mcp"
107
- }
108
- ```
109
-
110
- ## 6. What to inspect in artifacts
111
-
112
- For a repo-backed attempt, look at:
113
-
114
- - `contract.json`
115
- - `state.json`
116
- - `ledger.jsonl`
117
- - `artifacts/attempt-XXX/compiled-context.json`
118
- - `artifacts/attempt-XXX/diff.patch`
119
- - `artifacts/attempt-XXX/grounding-scan.json`
120
- - `artifacts/attempt-XXX/leash.json`
121
- - `artifacts/attempt-XXX/patch-score.json`
122
- - `artifacts/attempt-XXX/patch-decision.json`
123
- - `artifacts/attempt-XXX/rollback-boundary.json`
124
- - `artifacts/attempt-XXX/rollback-outcome.json`
125
-
126
- Those files are the evidence trail that backs the runtime’s claims.
1
+ # Examples
2
+
3
+ These examples are grounded in the current CLI and MCP surfaces in this repo. Where an example depends on a real provider path, it is labeled that way explicitly.
4
+
5
+ These are still primarily repo-local RC examples. The root `martin-loop` package facade is now real and smoke-validated, but registry publication remains a later release step.
6
+
7
+ ## 1. Stub-backed hello world
8
+
9
+ Use this when you want a safe first pass through the loop without real model spend.
10
+
11
+ ### PowerShell
12
+
13
+ ```powershell
14
+ $env:MARTIN_LIVE='false'
15
+ pnpm run:cli -- run `
16
+ --workspace ws_demo `
17
+ --project proj_demo `
18
+ --objective "Describe the current Martin run lifecycle in one paragraph" `
19
+ --verify "pnpm --filter @martin/core test"
20
+ Remove-Item Env:MARTIN_LIVE
21
+ ```
22
+
23
+ Why this is useful:
24
+
25
+ - exercises `runMartin`
26
+ - writes a real loop record and artifacts
27
+ - avoids external provider dependencies
28
+
29
+ ## 2. Repo-backed task with explicit scope
30
+
31
+ Use allow and deny paths so the task contract is narrow and reviewable.
32
+
33
+ ```bash
34
+ pnpm run:cli -- run \
35
+ --cwd . \
36
+ --objective "Tighten README wording for the OSS quickstart" \
37
+ --verify "pnpm --filter @martin/core test" \
38
+ --allow-path README.md \
39
+ --allow-path docs/oss/** \
40
+ --deny-path apps/control-plane/** \
41
+ --accept "Only update documentation files" \
42
+ --accept "Do not modify runtime code"
43
+ ```
44
+
45
+ What this demonstrates:
46
+
47
+ - repo root selection with `--cwd`
48
+ - scoped file-edit boundaries
49
+ - acceptance criteria injection into the task contract
50
+
51
+ ## 3. Safety-block example
52
+
53
+ This example is expected to block before execution because the verifier command is unsafe.
54
+
55
+ ```bash
56
+ pnpm run:cli -- run \
57
+ --objective "Try to run an unsafe verifier" \
58
+ --verify "rm -rf ."
59
+ ```
60
+
61
+ Expected behavior:
62
+
63
+ - the leash blocks the verifier command before adapter execution
64
+ - the run exits through a safety-oriented path rather than pretending the command was acceptable
65
+ - the attempt artifact set includes a persisted leash artifact when applicable
66
+
67
+ The point of this example is not that `rm` exists on every machine. The point is that the raw verifier text is evaluated before the process would be allowed to run.
68
+
69
+ ## 4. Budget-constrained live run
70
+
71
+ This is a live-provider example. Only use it when you have the relevant CLI and credentials configured.
72
+
73
+ ```bash
74
+ pnpm run:cli -- run \
75
+ --engine codex \
76
+ --model o3 \
77
+ --objective "Refactor the CLI argument parser for clarity" \
78
+ --verify "pnpm --filter @martin/cli test" \
79
+ --budget-usd 2 \
80
+ --soft-limit-usd 1 \
81
+ --max-iterations 2
82
+ ```
83
+
84
+ What to review afterward:
85
+
86
+ - admission and settlement events in `ledger.jsonl`
87
+ - cost provenance labels in the run artifacts
88
+ - whether the loop stopped for completion, budget pressure, or lack of progress
89
+
90
+ ## 5. MCP invocation shape
91
+
92
+ The MCP server exposes `martin_run`, `martin_inspect`, and `martin_status`.
93
+
94
+ Example `martin_run` payload:
95
+
96
+ ```json
97
+ {
98
+ "objective": "Tighten the local dashboard copy",
99
+ "workingDirectory": ".",
100
+ "engine": "claude",
101
+ "verificationPlan": ["pnpm --filter @martin/control-plane test"],
102
+ "maxUsd": 5,
103
+ "maxIterations": 2,
104
+ "maxTokens": 20000,
105
+ "workspaceId": "ws_mcp",
106
+ "projectId": "proj_mcp"
107
+ }
108
+ ```
109
+
110
+ ## 6. What to inspect in artifacts
111
+
112
+ For a repo-backed attempt, look at:
113
+
114
+ - `contract.json`
115
+ - `state.json`
116
+ - `ledger.jsonl`
117
+ - `artifacts/attempt-XXX/compiled-context.json`
118
+ - `artifacts/attempt-XXX/diff.patch`
119
+ - `artifacts/attempt-XXX/grounding-scan.json`
120
+ - `artifacts/attempt-XXX/leash.json`
121
+ - `artifacts/attempt-XXX/patch-score.json`
122
+ - `artifacts/attempt-XXX/patch-decision.json`
123
+ - `artifacts/attempt-XXX/rollback-boundary.json`
124
+ - `artifacts/attempt-XXX/rollback-outcome.json`
125
+
126
+ Those files are the evidence trail that backs the runtime’s claims.
@@ -1,113 +1,113 @@
1
- {
2
- "generatedAt": "2026-04-22T11:35:54.851Z",
3
- "verdict": "go",
4
- "publicSurface": {
5
- "packageName": "martin-loop",
6
- "canonicalPackageManager": "npm",
7
- "installCommand": "npm install martin-loop",
8
- "npxCommand": "npx martin-loop",
9
- "sdkImportPath": "martin-loop",
10
- "supportsNpxCommand": true,
11
- "supportsSdkImport": true
12
- },
13
- "ossCorePackages": [
14
- {
15
- "name": "@martin/contracts",
16
- "path": "packages/contracts",
17
- "private": true,
18
- "publishAccess": null,
19
- "workspaceDependencies": [],
20
- "classification": "oss_core",
21
- "classificationReason": "Intended Phase 13 OSS core surface."
22
- },
23
- {
24
- "name": "@martin/core",
25
- "path": "packages/core",
26
- "private": true,
27
- "publishAccess": null,
28
- "workspaceDependencies": [
29
- "@martin/contracts"
30
- ],
31
- "classification": "oss_core",
32
- "classificationReason": "Intended Phase 13 OSS core surface."
33
- },
34
- {
35
- "name": "@martin/adapters",
36
- "path": "packages/adapters",
37
- "private": true,
38
- "publishAccess": null,
39
- "workspaceDependencies": [
40
- "@martin/core"
41
- ],
42
- "classification": "oss_core",
43
- "classificationReason": "Intended Phase 13 OSS core surface."
44
- },
45
- {
46
- "name": "@martin/cli",
47
- "path": "packages/cli",
48
- "private": false,
49
- "publishAccess": "public",
50
- "workspaceDependencies": [
51
- "@martin/adapters",
52
- "@martin/contracts",
53
- "@martin/core"
54
- ],
55
- "classification": "oss_core",
56
- "classificationReason": "Intended Phase 13 OSS core surface."
57
- },
58
- {
59
- "name": "@martin/mcp",
60
- "path": "packages/mcp",
61
- "private": false,
62
- "publishAccess": "public",
63
- "workspaceDependencies": [
64
- "@martin/adapters",
65
- "@martin/contracts",
66
- "@martin/core"
67
- ],
68
- "classification": "oss_core",
69
- "classificationReason": "Intended Phase 13 OSS core surface."
70
- }
71
- ],
72
- "nonOssWorkspacePackages": [
73
- {
74
- "name": "@martin/control-plane",
75
- "path": "apps/control-plane",
76
- "private": true,
77
- "publishAccess": null,
78
- "workspaceDependencies": [
79
- "@martin/contracts"
80
- ],
81
- "classification": "non_oss_workspace",
82
- "classificationReason": "Managed or RC-only workspace surface that stays out of the initial OSS boundary."
83
- },
84
- {
85
- "name": "@martin/benchmarks",
86
- "path": "benchmarks",
87
- "private": true,
88
- "publishAccess": null,
89
- "workspaceDependencies": [
90
- "@martin/adapters",
91
- "@martin/contracts",
92
- "@martin/core"
93
- ],
94
- "classification": "non_oss_workspace",
95
- "classificationReason": "Managed or RC-only workspace surface that stays out of the initial OSS boundary."
96
- }
97
- ],
98
- "localOnlySurfaces": [
99
- {
100
- "path": "apps/local-dashboard",
101
- "reason": "Local read-model viewer that is not yet packaged as a publishable OSS workspace."
102
- }
103
- ],
104
- "dependencyLeaks": [],
105
- "summary": {
106
- "ossCoreCount": 5,
107
- "nonOssWorkspaceCount": 2,
108
- "localOnlySurfaceCount": 1,
109
- "dependencyLeakCount": 0,
110
- "privateOssCoreCount": 3,
111
- "publishReadyOssCoreCount": 2
112
- }
113
- }
1
+ {
2
+ "generatedAt": "2026-04-23T15:03:09.849Z",
3
+ "verdict": "go",
4
+ "publicSurface": {
5
+ "packageName": "martin-loop",
6
+ "canonicalPackageManager": "npm",
7
+ "installCommand": "npm install martin-loop",
8
+ "npxCommand": "npx martin-loop",
9
+ "sdkImportPath": "martin-loop",
10
+ "supportsNpxCommand": true,
11
+ "supportsSdkImport": true
12
+ },
13
+ "ossCorePackages": [
14
+ {
15
+ "name": "@martin/contracts",
16
+ "path": "packages/contracts",
17
+ "private": true,
18
+ "publishAccess": null,
19
+ "workspaceDependencies": [],
20
+ "classification": "oss_core",
21
+ "classificationReason": "Intended Phase 13 OSS core surface."
22
+ },
23
+ {
24
+ "name": "@martin/core",
25
+ "path": "packages/core",
26
+ "private": true,
27
+ "publishAccess": null,
28
+ "workspaceDependencies": [
29
+ "@martin/contracts"
30
+ ],
31
+ "classification": "oss_core",
32
+ "classificationReason": "Intended Phase 13 OSS core surface."
33
+ },
34
+ {
35
+ "name": "@martin/adapters",
36
+ "path": "packages/adapters",
37
+ "private": true,
38
+ "publishAccess": null,
39
+ "workspaceDependencies": [
40
+ "@martin/core"
41
+ ],
42
+ "classification": "oss_core",
43
+ "classificationReason": "Intended Phase 13 OSS core surface."
44
+ },
45
+ {
46
+ "name": "@martin/cli",
47
+ "path": "packages/cli",
48
+ "private": false,
49
+ "publishAccess": "public",
50
+ "workspaceDependencies": [
51
+ "@martin/adapters",
52
+ "@martin/contracts",
53
+ "@martin/core"
54
+ ],
55
+ "classification": "oss_core",
56
+ "classificationReason": "Intended Phase 13 OSS core surface."
57
+ },
58
+ {
59
+ "name": "@martin/mcp",
60
+ "path": "packages/mcp",
61
+ "private": false,
62
+ "publishAccess": "public",
63
+ "workspaceDependencies": [
64
+ "@martin/adapters",
65
+ "@martin/contracts",
66
+ "@martin/core"
67
+ ],
68
+ "classification": "oss_core",
69
+ "classificationReason": "Intended Phase 13 OSS core surface."
70
+ }
71
+ ],
72
+ "nonOssWorkspacePackages": [
73
+ {
74
+ "name": "@martin/control-plane",
75
+ "path": "apps/control-plane",
76
+ "private": true,
77
+ "publishAccess": null,
78
+ "workspaceDependencies": [
79
+ "@martin/contracts"
80
+ ],
81
+ "classification": "non_oss_workspace",
82
+ "classificationReason": "Managed or RC-only workspace surface that stays out of the initial OSS boundary."
83
+ },
84
+ {
85
+ "name": "@martin/benchmarks",
86
+ "path": "benchmarks",
87
+ "private": true,
88
+ "publishAccess": null,
89
+ "workspaceDependencies": [
90
+ "@martin/adapters",
91
+ "@martin/contracts",
92
+ "@martin/core"
93
+ ],
94
+ "classification": "non_oss_workspace",
95
+ "classificationReason": "Managed or RC-only workspace surface that stays out of the initial OSS boundary."
96
+ }
97
+ ],
98
+ "localOnlySurfaces": [
99
+ {
100
+ "path": "apps/local-dashboard",
101
+ "reason": "Local read-model viewer that is not yet packaged as a publishable OSS workspace."
102
+ }
103
+ ],
104
+ "dependencyLeaks": [],
105
+ "summary": {
106
+ "ossCoreCount": 5,
107
+ "nonOssWorkspaceCount": 2,
108
+ "localOnlySurfaceCount": 1,
109
+ "dependencyLeakCount": 0,
110
+ "privateOssCoreCount": 3,
111
+ "publishReadyOssCoreCount": 2
112
+ }
113
+ }
@@ -1,48 +1,48 @@
1
- # Martin Loop Phase 13 OSS Core Boundary
2
-
3
- Generated: 2026-04-22T11:35:54.851Z
4
-
5
- ## Verdict
6
- **GO**
7
-
8
- ## Summary
9
- - Public package target: martin-loop
10
- - Canonical public package manager: npm
11
- - Intended OSS core packages: 5
12
- - Non-OSS workspace packages: 2
13
- - Local-only surfaces: 1
14
- - Private OSS-core packages still gated from publish: 3
15
- - OSS-core packages already publish-configured: 2
16
- - Dependency leaks: 0
17
- - No workspace dependency leaks detected between the intended OSS core and the non-OSS workspace surfaces.
18
-
19
- ## Public Package Surface
20
- - Install target: `npm install martin-loop`
21
- - CLI target: `npx martin-loop`
22
- - SDK target: `import { MartinLoop } from 'martin-loop'`
23
- - Root `npx martin-loop` support shipped: yes
24
- - Root SDK import shipped: yes
25
-
26
- ## Intended OSS Core Packages
27
- | Package | Path | Private | Publish Access | Workspace Deps |
28
- |---|---|---|---|---|
29
- | @martin/contracts | packages/contracts | yes | n/a | none |
30
- | @martin/core | packages/core | yes | n/a | @martin/contracts |
31
- | @martin/adapters | packages/adapters | yes | n/a | @martin/core |
32
- | @martin/cli | packages/cli | no | public | @martin/adapters, @martin/contracts, @martin/core |
33
- | @martin/mcp | packages/mcp | no | public | @martin/adapters, @martin/contracts, @martin/core |
34
-
35
- ## Non-OSS Workspace Packages
36
- | Package | Path | Reason |
37
- |---|---|---|
38
- | @martin/control-plane | apps/control-plane | Managed or RC-only workspace surface that stays out of the initial OSS boundary. |
39
- | @martin/benchmarks | benchmarks | Managed or RC-only workspace surface that stays out of the initial OSS boundary. |
40
-
41
- ## Local-Only Surfaces
42
- | Path | Reason |
43
- |---|---|
44
- | apps/local-dashboard | Local read-model viewer that is not yet packaged as a publishable OSS workspace. |
45
-
46
- ## Dependency Leak Review
47
- - No workspace dependency leaks detected.
48
-
1
+ # Martin Loop Phase 13 OSS Core Boundary
2
+
3
+ Generated: 2026-04-23T15:03:09.849Z
4
+
5
+ ## Verdict
6
+ **GO**
7
+
8
+ ## Summary
9
+ - Public package target: martin-loop
10
+ - Canonical public package manager: npm
11
+ - Intended OSS core packages: 5
12
+ - Non-OSS workspace packages: 2
13
+ - Local-only surfaces: 1
14
+ - Private OSS-core packages still gated from publish: 3
15
+ - OSS-core packages already publish-configured: 2
16
+ - Dependency leaks: 0
17
+ - No workspace dependency leaks detected between the intended OSS core and the non-OSS workspace surfaces.
18
+
19
+ ## Public Package Surface
20
+ - Install target: `npm install martin-loop`
21
+ - CLI target: `npx martin-loop`
22
+ - SDK target: `import { MartinLoop } from 'martin-loop'`
23
+ - Root `npx martin-loop` support shipped: yes
24
+ - Root SDK import shipped: yes
25
+
26
+ ## Intended OSS Core Packages
27
+ | Package | Path | Private | Publish Access | Workspace Deps |
28
+ |---|---|---|---|---|
29
+ | @martin/contracts | packages/contracts | yes | n/a | none |
30
+ | @martin/core | packages/core | yes | n/a | @martin/contracts |
31
+ | @martin/adapters | packages/adapters | yes | n/a | @martin/core |
32
+ | @martin/cli | packages/cli | no | public | @martin/adapters, @martin/contracts, @martin/core |
33
+ | @martin/mcp | packages/mcp | no | public | @martin/adapters, @martin/contracts, @martin/core |
34
+
35
+ ## Non-OSS Workspace Packages
36
+ | Package | Path | Reason |
37
+ |---|---|---|
38
+ | @martin/control-plane | apps/control-plane | Managed or RC-only workspace surface that stays out of the initial OSS boundary. |
39
+ | @martin/benchmarks | benchmarks | Managed or RC-only workspace surface that stays out of the initial OSS boundary. |
40
+
41
+ ## Local-Only Surfaces
42
+ | Path | Reason |
43
+ |---|---|
44
+ | apps/local-dashboard | Local read-model viewer that is not yet packaged as a publishable OSS workspace. |
45
+
46
+ ## Dependency Leak Review
47
+ - No workspace dependency leaks detected.
48
+