edify-agents-cli 0.2.1__tar.gz → 0.3.0__tar.gz

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 (168) hide show
  1. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/.gitignore +4 -0
  2. edify_agents_cli-0.3.0/CHANGELOG.md +142 -0
  3. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/INSTALL.md +8 -44
  4. edify_agents_cli-0.3.0/PKG-INFO +276 -0
  5. edify_agents_cli-0.3.0/README.md +227 -0
  6. edify_agents_cli-0.3.0/edify/commands/plan.md +33 -0
  7. edify_agents_cli-0.3.0/edify/commands/spec.md +31 -0
  8. edify_agents_cli-0.3.0/edify/commands/tasks.md +47 -0
  9. edify_agents_cli-0.3.0/edify/commands/verify.md +46 -0
  10. edify_agents_cli-0.3.0/edify/formats/plan.example.md +108 -0
  11. edify_agents_cli-0.3.0/edify/formats/plan.format.md +40 -0
  12. edify_agents_cli-0.3.0/edify/formats/spec.example.md +183 -0
  13. edify_agents_cli-0.3.0/edify/formats/spec.format.md +42 -0
  14. edify_agents_cli-0.3.0/edify/formats/tasks.example.md +423 -0
  15. edify_agents_cli-0.3.0/edify/formats/tasks.format.md +84 -0
  16. edify_agents_cli-0.3.0/edify/manifest.md +26 -0
  17. edify_agents_cli-0.3.0/edify/mcp/default-registry.md +30 -0
  18. edify_agents_cli-0.3.0/edify/skills/adversary.md +39 -0
  19. edify_agents_cli-0.3.0/edify/skills/debugger.md +44 -0
  20. edify_agents_cli-0.3.0/edify/skills/implementer-general.md +46 -0
  21. edify_agents_cli-0.3.0/edify/skills/implementer-go.md +35 -0
  22. edify_agents_cli-0.3.0/edify/skills/implementer-node.md +35 -0
  23. edify_agents_cli-0.3.0/edify/skills/implementer-postgres.md +37 -0
  24. edify_agents_cli-0.3.0/edify/skills/implementer-python.md +36 -0
  25. edify_agents_cli-0.3.0/edify/skills/implementer-rails.md +49 -0
  26. edify_agents_cli-0.3.0/edify/skills/implementer-react.md +36 -0
  27. edify_agents_cli-0.3.0/edify/skills/planner.md +35 -0
  28. edify_agents_cli-0.3.0/edify/skills/test-writer-node.md +47 -0
  29. edify_agents_cli-0.3.0/edify/skills/test-writer-python.md +49 -0
  30. edify_agents_cli-0.3.0/edify/skills/verifier.md +40 -0
  31. edify_agents_cli-0.3.0/edify/templates/CLAUDE.md +19 -0
  32. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/__init__.py +1 -1
  33. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/agents.py +6 -5
  34. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/cli.py +277 -311
  35. edify_agents_cli-0.3.0/src/edify/commands/doctor_cmd.py +364 -0
  36. edify_agents_cli-0.3.0/src/edify/commands/graph_cmd.py +344 -0
  37. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/init_cmd.py +1 -0
  38. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/mcp_cmd.py +45 -12
  39. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/skills_cmd.py +1 -1
  40. edify_agents_cli-0.3.0/src/edify/commands/swarm_cmd.py +369 -0
  41. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/context.py +9 -3
  42. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/errors.py +28 -8
  43. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/mcp.py +9 -0
  44. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/tasks.py +2 -2
  45. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/extract.py +64 -8
  46. edify_agents_cli-0.3.0/src/edify/graph/insight.py +582 -0
  47. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/langs/python_ast.py +43 -17
  48. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/model.py +5 -4
  49. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/queries.py +46 -27
  50. edify_agents_cli-0.3.0/src/edify/graph/store.py +286 -0
  51. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/host.py +26 -9
  52. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/paths.py +12 -0
  53. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/skills.py +1 -1
  54. edify_agents_cli-0.3.0/src/edify/swarm.py +585 -0
  55. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/ui.py +60 -5
  56. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/conftest.py +5 -0
  57. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_cli.py +212 -0
  58. edify_agents_cli-0.3.0/tests/test_graph_paging.py +92 -0
  59. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_harvest.py +10 -0
  60. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_init_new.py +1 -1
  61. edify_agents_cli-0.3.0/tests/test_insight.py +70 -0
  62. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_library.py +15 -1
  63. edify_agents_cli-0.3.0/tests/test_swarm.py +176 -0
  64. edify_agents_cli-0.3.0/tests/test_swarm_handback.py +190 -0
  65. edify_agents_cli-0.3.0/tests/test_swarm_merge.py +132 -0
  66. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_ui.py +47 -0
  67. edify_agents_cli-0.2.1/CHANGELOG.md +0 -433
  68. edify_agents_cli-0.2.1/PKG-INFO +0 -411
  69. edify_agents_cli-0.2.1/README.md +0 -362
  70. edify_agents_cli-0.2.1/edify/commands/build.md +0 -64
  71. edify_agents_cli-0.2.1/edify/commands/plan.md +0 -68
  72. edify_agents_cli-0.2.1/edify/commands/spec.md +0 -63
  73. edify_agents_cli-0.2.1/edify/commands/tasks.md +0 -65
  74. edify_agents_cli-0.2.1/edify/commands/verify.md +0 -67
  75. edify_agents_cli-0.2.1/edify/formats/plan.example.md +0 -455
  76. edify_agents_cli-0.2.1/edify/formats/plan.format.md +0 -128
  77. edify_agents_cli-0.2.1/edify/formats/spec.example.md +0 -483
  78. edify_agents_cli-0.2.1/edify/formats/spec.format.md +0 -116
  79. edify_agents_cli-0.2.1/edify/formats/tasks.example.md +0 -712
  80. edify_agents_cli-0.2.1/edify/formats/tasks.format.md +0 -215
  81. edify_agents_cli-0.2.1/edify/manifest.md +0 -30
  82. edify_agents_cli-0.2.1/edify/mcp/default-registry.md +0 -66
  83. edify_agents_cli-0.2.1/edify/skills/adversary.md +0 -61
  84. edify_agents_cli-0.2.1/edify/skills/debugger.md +0 -64
  85. edify_agents_cli-0.2.1/edify/skills/implementer-general.md +0 -53
  86. edify_agents_cli-0.2.1/edify/skills/implementer-go.md +0 -63
  87. edify_agents_cli-0.2.1/edify/skills/implementer-node.md +0 -61
  88. edify_agents_cli-0.2.1/edify/skills/implementer-postgres.md +0 -63
  89. edify_agents_cli-0.2.1/edify/skills/implementer-python.md +0 -65
  90. edify_agents_cli-0.2.1/edify/skills/implementer-react.md +0 -65
  91. edify_agents_cli-0.2.1/edify/skills/planner.md +0 -58
  92. edify_agents_cli-0.2.1/edify/skills/test-writer-node.md +0 -64
  93. edify_agents_cli-0.2.1/edify/skills/test-writer-python.md +0 -60
  94. edify_agents_cli-0.2.1/edify/skills/verifier.md +0 -67
  95. edify_agents_cli-0.2.1/edify/templates/CLAUDE.md +0 -18
  96. edify_agents_cli-0.2.1/src/edify/commands/doctor_cmd.py +0 -206
  97. edify_agents_cli-0.2.1/src/edify/commands/graph_cmd.py +0 -156
  98. edify_agents_cli-0.2.1/src/edify/graph/store.py +0 -157
  99. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/LICENSE +0 -0
  100. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/edify/harvest/harvest.md +0 -0
  101. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/edify/harvest/sources.md +0 -0
  102. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/edify/harvest/vocab.md +0 -0
  103. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/pyproject.toml +0 -0
  104. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/__main__.py +0 -0
  105. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/_toml.py +0 -0
  106. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/anim.py +0 -0
  107. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/assets.py +0 -0
  108. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/__init__.py +0 -0
  109. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/check_cmd.py +0 -0
  110. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/feedback_cmd.py +0 -0
  111. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/governance_cmd.py +0 -0
  112. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/harvest_cmd.py +0 -0
  113. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/init_new_cmd.py +0 -0
  114. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/license_cmd.py +0 -0
  115. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/self_cmd.py +0 -0
  116. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/setup_cmd.py +0 -0
  117. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/commands/upgrade_cmd.py +0 -0
  118. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/conventions.py +0 -0
  119. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/detect.py +0 -0
  120. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/distribution.py +0 -0
  121. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/feedback.py +0 -0
  122. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/__init__.py +0 -0
  123. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/check.py +0 -0
  124. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/document.py +0 -0
  125. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/finding.py +0 -0
  126. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/fix.py +0 -0
  127. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/plan.py +0 -0
  128. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/skill.py +0 -0
  129. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/formats/spec.py +0 -0
  130. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/governance.py +0 -0
  131. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/__init__.py +0 -0
  132. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/ignore.py +0 -0
  133. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/langs/__init__.py +0 -0
  134. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/langs/ctags.py +0 -0
  135. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/langs/markdown.py +0 -0
  136. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/langs/patterns.py +0 -0
  137. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/graph/langs/sqlish.py +0 -0
  138. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/harvest/__init__.py +0 -0
  139. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/harvest/ledger.py +0 -0
  140. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/harvest/screen.py +0 -0
  141. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/harvest/sources.py +0 -0
  142. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/interview.py +0 -0
  143. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/library.py +0 -0
  144. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/__init__.py +0 -0
  145. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/binding.py +0 -0
  146. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/ed25519.py +0 -0
  147. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/projects.py +0 -0
  148. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/revocation.py +0 -0
  149. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/tier.py +0 -0
  150. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/licensing/token.py +0 -0
  151. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/selfinstall.py +0 -0
  152. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/src/edify/tsv.py +0 -0
  153. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_anim.py +0 -0
  154. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_detect.py +0 -0
  155. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_feedback.py +0 -0
  156. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_fix.py +0 -0
  157. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_formats.py +0 -0
  158. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_governance.py +0 -0
  159. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_graph.py +0 -0
  160. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_host.py +0 -0
  161. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_license_cmd.py +0 -0
  162. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_licensing.py +0 -0
  163. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_mixed_versions.py +0 -0
  164. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_platform.py +0 -0
  165. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_projects_retired.py +0 -0
  166. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_selection.py +0 -0
  167. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_self.py +0 -0
  168. {edify_agents_cli-0.2.1 → edify_agents_cli-0.3.0}/tests/test_toml.py +0 -0
@@ -29,3 +29,7 @@ edify_public/
29
29
  # SSH deploy keys never belong in the repository
30
30
  edify_deploy
31
31
  edify_deploy.pub
32
+ .edify/run/
33
+
34
+ # Agent worktrees — each one is its own checkout
35
+ .claude/worktrees/
@@ -0,0 +1,142 @@
1
+ # Changelog
2
+
3
+ All notable changes to `edify-agents-cli`. The format follows Keep a Changelog; versioning
4
+ is semantic on the CLI surface.
5
+
6
+ Two contract versions move independently of the CLI version and are recorded in
7
+ `.edify/graph/meta` and `edify/manifest.md`: `graph-schema` (bumping it rebuilds an
8
+ existing graph rather than misreading it) and `index-schema`.
9
+
10
+ ## 0.3.0
11
+
12
+ Built for huge repositories, swarms of agents, and several people on one feature.
13
+
14
+ - **Four commands.** `/spec`, `/plan`, `/tasks`, `/verify`, each 30–50 lines. `/build`
15
+ folds into `/tasks`: write the task list, then run it with one loop. Unknowns are
16
+ marked `[NEEDS CLARIFICATION: …]`. `/verify` exercises every surface a spec declares
17
+ (service, browser, device or emulator, live API keys, connectors) and reports a tier it
18
+ cannot run as `blocked: needs X`, never as a pass. Format contracts are a third of their
19
+ former length.
20
+ - **The graph summarises itself.** `edify graph map` writes `.edify/graph/MAP.md`:
21
+ communities (deterministic label propagation over the file graph), hubs and bridges.
22
+ `graph context "<task>"` returns the slice one task needs under a token budget.
23
+ `graph explain` and `graph path` tag each edge EXTRACTED or INFERRED and exit 0
24
+ (answered), 3 (ambiguous, pin with `file::name`) or 4 (not found).
25
+ - **Incremental graph.** `graph update --changed` re-extracts only what git reports.
26
+ A scoped update now resolves references out of the re-extracted directories instead of
27
+ dropping them. `graph hook` installs git hooks that keep the graph current.
28
+ - **Swarm leases.** `edify swarm board|claim|prompt|done|block|release|renew`. Claims are
29
+ gated on phase and dependencies, and a claim waits while its blast radius (files plus
30
+ graph dependents) touches a live lease. Leases live in the git common dir by default.
31
+ With `--git` they live on `origin/edify-swarm`, where push atomicity is the lock. `done`
32
+ is fenced: an expired holder cannot record over the new one. `swarm prompt` is the
33
+ whole spawn payload: resolved skill, task block, decisions, graph slice, scoped servers.
34
+ - **MCP and plugins only when needed.** `edify mcp config <role> <phase>` writes only the
35
+ scoped servers, from `.edify/mcp.pool.json` (no runtime auto-loads it), for
36
+ `--mcp-config … --strict-mcp-config`.
37
+ - **Library.** The harvested methods are folded into the role skills that resolve
38
+ automatically: verification-gap and edge-case-trace into `verifier`, submit-gate and
39
+ evidence-before-claim into `implementer-general`, reproduction-first and
40
+ network-diagnosis into `debugger`, finding-triage into `adversary`, and test-critique
41
+ and condition-based-waiting into both test writers. `implementer-rails` is new. Role
42
+ skills are 23–33 lines.
43
+
44
+ ## 0.2.1
45
+
46
+ - **Licences carry the team fields.** A team or partner token names its client, its admin
47
+ handles, an optional repository binding (the root commit) and its signing key (`kid`).
48
+ Every field is optional, so tokens already issued keep working.
49
+ - **Licence ID.** Every licence has one, derived from its signed bytes, shown by
50
+ `edify license status` and in `--json` as `license.id`.
51
+ - **Revocation, offline.** A release can carry a signed list of revoked licence IDs
52
+ (`edify/licensing/revoked.tsv`). A revoked licence runs the public edition and says so;
53
+ `activate` refuses it. Nothing is fetched.
54
+ - **Key rotation.** Several licence keys can ship at once, selected by the token's `kid`.
55
+ `EDIFY_LICENSE_PUBKEY` now adds a key rather than replacing the shipped one.
56
+ - `license status` shows the client, admins, licence ID, verifying key and whether a bound
57
+ licence matches this repository. `status` and `doctor` warn 30 days before expiry
58
+ (was 15).
59
+ - `edify check` understands milestone-scoped task lists. With `milestone:` in the
60
+ frontmatter, a `deferred` coverage row excuses the assertions it owns (listed as notes,
61
+ not errors), and a delete by a task marked `Status — done` no longer counts as a
62
+ missing file. `deferred` without a milestone is an error.
63
+ - Fixed: `edify license activate <token>` crashed with "File name too long" when the
64
+ token, pasted on the command line, was longer than a filename may be.
65
+ - Fixed: a validly signed token with a non-numeric `seats`, or a payload that is not an
66
+ object, crashed licence resolution instead of being refused.
67
+
68
+ ## 0.2.0
69
+
70
+ - **Renamed** on PyPI from `edify-cli`, a name that belongs to another project, to
71
+ `edify-agents-cli`.
72
+ - Removed: the individual paid plan (`pro`), the purchase and project-count
73
+ subcommands, and the graph, project and upgrade caps.
74
+ - Changed: a second MCP server needs a team or partner licence, and existing entries are
75
+ kept.
76
+ - Docs: the README and `docs/` now carry the site's ink clips as animated WebP
77
+ (`assets/motion/`), and the terminal screenshots are recaptured from 0.2.0 output.
78
+
79
+ ## Between 0.1.0 and 0.2.0
80
+
81
+ Condensed; the full notes are in git history (`git log -p -- CHANGELOG.md`).
82
+
83
+ - **A real issuing key.** 0.1.0 shipped the public key of an all-zeros seed; it was
84
+ replaced with a random key whose private half never enters this repository, and
85
+ `tests/test_licensing.py` fails if a trivially derivable key comes back.
86
+ - **`edify self update [--from PATH]` and `edify self where`.** Replaces the CLI through
87
+ whichever tool installed it (uv, pipx or pip; `--manager` overrides detection),
88
+ `--dry-run` prints the exact argv, a locked `edify.exe` on Windows is reported rather
89
+ than hidden, and `--offline` refuses to fetch `hatchling`. Never touches `.edify/`.
90
+ - **The install has a face.** A wordmark and an animated ∞ (a lemniscate of Gerono,
91
+ precomputed frames, two glyph sets by stream encoding) paint to stderr on a real
92
+ terminal only; `--json` output is unchanged and `EDIFY_NO_ANIM` turns it off.
93
+ - **`edify init new [PATH]`.** A folder set up from zero for every runtime in one table
94
+ (`src/edify/agents.py`): Claude Code, Codex, Cursor, Gemini CLI, Copilot, Windsurf and
95
+ a generic `AGENTS.md`. Eight skippable questions write `.edify/profile.md`. Clearing is
96
+ one implementation (`host.clear`) shared with `setup --fresh`, and a later `init`
97
+ keeps every runtime already set up current.
98
+ - **`edify setup [PATH] [--fresh]`.** Pins the root where pointed instead of walking up
99
+ into a parent checkout; `--fresh` clears EDIFY's own files and nothing else.
100
+ - **Installed skills are visible to the runtime.** `edify skills sync` writes pointer
101
+ files at `.claude/skills/<name>/SKILL.md`; a hand-authored `SKILL.md` is never touched.
102
+ - **`/harvest`.** `edify harvest pick | stage | screen | propose | admit | reject |
103
+ status`, with `harvest/ledger.json` recognising a candidate by URL, content hash and
104
+ entry name. `pick` scores sources deterministically; the screen (licence allowlist,
105
+ duplicate, injection scan) fails closed; `admit` needs a person at a terminal and then
106
+ installs, indexes, mirrors and records governance in one step.
107
+ - **Governance over every installed file.** `.edify/governance.tsv` and
108
+ `edify governance list | verify | rebuild`; `verify` tells a modified EDIFY file from
109
+ an edited repository-owned one.
110
+ - **The library asks before it installs** (`--skills`, `--yes`); with no terminal it asks
111
+ nothing.
112
+ - **Feedback with no telemetry.** `edify feedback` writes four answers to a local file
113
+ and prints how to send them; nothing is transmitted.
114
+ - **macOS and publishing.** Platform config directories, a walk that ignores macOS
115
+ bundle debris, `INSTALL.md`, the Homebrew formula, `PUBLISHING.md`, and a tag-driven
116
+ `release.yml` with trusted publishing and wheel smoke tests on three platforms.
117
+ - Fixed: a `→` or `·` in output no longer kills a command on a cp1252 Windows console.
118
+ - A paid individual plan and its project, graph and upgrade caps were added here and
119
+ removed in 0.2.0.
120
+
121
+ ## [0.1.0] — 2026-08-07
122
+
123
+ First release: the graph commands, the methodology surface, and the install path.
124
+
125
+ - **The map.** `edify graph build | update | where | dependents | defines | overlap |
126
+ references | stat`, deterministic and byte-identical on unchanged source. Python is
127
+ parsed with a real syntax tree, 19 languages are scanned, and `meta.method_<language>`
128
+ and `meta.languages_uncovered` say which. No model writes the graph. Optional
129
+ `--extractor ctags`.
130
+ - **The documents.** `edify check [--fix] [--exit-code]` over specs, plans, tasks, skill
131
+ files and the server registry; it reports and never blocks.
132
+ - **The install.** `edify init`, `skills list | resolve | index | add`,
133
+ `mcp list | check | for`, `doctor`, and `upgrade` (with `--archive` for air-gapped
134
+ machines).
135
+ - **The methodology tree.** Five command files (`/spec`, `/plan`, `/tasks`, `/build`,
136
+ `/verify`; `/build` folded into `/tasks` in 0.3.0), three format contracts with worked
137
+ examples, twelve skill entries, the harvest design, and a one-entry registry.
138
+ - **Licensing.** Offline Ed25519 verification in pure Python (RFC 8032 §7.1 vector),
139
+ `edify license status | activate | deactivate`, and the issuer.
140
+ - **Known limits.** Everything but Python, Markdown and SQL is scanned, not parsed
141
+ (`doctor` says amber); `mcp check` cannot probe remote or version-less servers; no
142
+ standalone binaries.
@@ -16,7 +16,7 @@ it installs one command, `edify`.
16
16
  ```bash
17
17
  pip install edify-agents-cli
18
18
 
19
- edify version # edify 0.2.1
19
+ edify version # edify 0.3.0
20
20
  ```
21
21
 
22
22
  `pip` installs into whichever Python environment it belongs to. Inside a virtual
@@ -141,40 +141,11 @@ sudo apt install universal-ctags # Debian/Ubuntu
141
141
  edify graph build --extractor ctags
142
142
  ```
143
143
 
144
- ## What `init` asks
144
+ ## What `init` asks, and installing for every agent
145
145
 
146
- One question: which entries from the agent library to install. They become trusted
147
- instruction in every matching session, so you see the list before any of it lands.
148
-
149
- With no terminal — CI, a pipe, a model calling the binary — nothing is asked and the
150
- whole stack-matched set is installed. To state the answer up front:
151
-
152
- ```bash
153
- edify init --yes # the whole matched set, no question
154
- edify init --skills none # none of it; add entries later
155
- edify init --skills planner,verifier # exactly these
156
- ```
157
-
158
- ## Setting a folder up from zero, for every agent
159
-
160
- `edify init` writes `CLAUDE.md` and `.claude/`. If your team also has Codex, Cursor,
161
- Gemini, Copilot, or Windsurf open, that install is invisible to them. `edify init new`
162
- pins the root where you point it, creates the folder if it is missing, clears EDIFY's
163
- own files first, installs every library entry, and writes the instruction file and
164
- command pointers each runtime actually reads.
165
-
166
- ```bash
167
- edify init new ./service # ask, then install for every runtime
168
- edify init new . --agents claude,cursor
169
- edify init new . --yes # ask nothing, install everything
170
- ```
171
-
172
- It asks eight short questions before installing anything — which runtimes, what stack
173
- detection missed, who you are, your role, what is being built, what done looks like, how
174
- much an agent should explain, and what it must never do here. Enter skips any of them.
175
- The answers land in `.edify/profile.md`, which every instruction file points at and
176
- which is yours to edit; a re-run offers them back as the defaults. With no terminal
177
- nothing is asked and no profile is written.
146
+ `init` asks one question, which library entries to install; `edify init new` asks eight
147
+ skippable ones and installs for every agent runtime. With no terminal, neither asks
148
+ anything. The flags and the runtime table are in the [README](README.md#install).
178
149
 
179
150
  ## Installing from a checkout
180
151
 
@@ -222,17 +193,10 @@ shell rather than reporting a success it did not have.
222
193
  `self update` never touches `.edify/` in any repository. Reinstalling the harness into a
223
194
  checkout is `edify setup . --fresh`.
224
195
 
225
- ## The public edition, and what a licence adds
226
-
227
- Nothing here needs an account. The public edition, `edify-agents-cli`, is MIT-licensed
228
- and free: every command, every query, every check, the whole methodology tree, on a
229
- repository of any size, in as many repositories as you like. It has no caps except one —
230
- the MCP registry scopes a single server into spawns.
196
+ ## Licences
231
197
 
232
- The team edition, `edify-agents-teams`, is by licence. A team or partner licence adds the
233
- team commands and lifts the one-server rule. `partner` plans are free for University of
234
- Toronto design teams and researchers. There is no payment and no price anywhere; a
235
- licence is issued on request.
198
+ The public edition needs no account and no licence; the [README](README.md#editions-privacy-and-what-edify-does-not-do)
199
+ says what a team licence adds.
236
200
 
237
201
  ```bash
238
202
  edify license status # which edition this machine runs, and what a licence adds
@@ -0,0 +1,276 @@
1
+ Metadata-Version: 2.5
2
+ Name: edify-agents-cli
3
+ Version: 0.3.0
4
+ Summary: A harness that makes an AI coding agent work reliably on large codebases: three documents and a map.
5
+ Project-URL: Homepage, https://github.com/EDIFY-agents/EDIFY
6
+ Project-URL: Documentation, https://github.com/EDIFY-agents/EDIFY#readme
7
+ Project-URL: Changelog, https://github.com/EDIFY-agents/EDIFY/blob/main/CHANGELOG.md
8
+ Project-URL: Issues, https://github.com/EDIFY-agents/EDIFY/issues
9
+ Author: EDIFY
10
+ License: MIT License
11
+
12
+ Copyright (c) 2026 Santiago
13
+
14
+ Permission is hereby granted, free of charge, to any person obtaining a copy
15
+ of this software and associated documentation files (the "Software"), to deal
16
+ in the Software without restriction, including without limitation the rights
17
+ to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
18
+ copies of the Software, and to permit persons to whom the Software is
19
+ furnished to do so, subject to the following conditions:
20
+
21
+ The above copyright notice and this permission notice shall be included in all
22
+ copies or substantial portions of the Software.
23
+
24
+ THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
25
+ IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
26
+ FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
27
+ AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
28
+ LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
29
+ OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
30
+ SOFTWARE.
31
+ License-File: LICENSE
32
+ Keywords: agent,ai,codebase,codegen,graph,harness,spec
33
+ Classifier: Development Status :: 4 - Beta
34
+ Classifier: Environment :: Console
35
+ Classifier: Intended Audience :: Developers
36
+ Classifier: Programming Language :: Python :: 3
37
+ Classifier: Programming Language :: Python :: 3.10
38
+ Classifier: Programming Language :: Python :: 3.11
39
+ Classifier: Programming Language :: Python :: 3.12
40
+ Classifier: Programming Language :: Python :: 3.13
41
+ Classifier: Topic :: Software Development :: Build Tools
42
+ Requires-Python: >=3.10
43
+ Provides-Extra: dev
44
+ Requires-Dist: pytest>=8.0; extra == 'dev'
45
+ Provides-Extra: treesitter
46
+ Requires-Dist: tree-sitter-language-pack>=0.7; extra == 'treesitter'
47
+ Requires-Dist: tree-sitter>=0.23; extra == 'treesitter'
48
+ Description-Content-Type: text/markdown
49
+
50
+ <div align="center">
51
+
52
+ <img src="assets/motion/stage.webp" width="100%" alt="The EDIFY wordmark, flat white type with two pools of blue ink travelling through it">
53
+
54
+ # EDIFY
55
+
56
+ **Three documents and a map. Think hard once, then build without thinking.**
57
+
58
+ [Install](#install) · [The four commands](#the-four-commands) · [The graph](#the-graph) · [Docs](docs/) · [Website](https://edify-agents.github.io/EDIFY/)
59
+
60
+ </div>
61
+
62
+ EDIFY is a harness that sits around an AI coding agent and makes it work reliably on
63
+ large, real codebases — a million lines, fifteen years of history, four languages. It
64
+ is not a model, not an agent framework, and not a wrapper around a chat interface: a
65
+ stdlib-only CLI, `edify`, plus four short commands your agent runs.
66
+
67
+ A frontier model with a plain setup writes excellent code and still fails on serious
68
+ projects, from four structural gaps no model generation closes. EDIFY closes exactly
69
+ those four and gets out of the way of everything else:
70
+
71
+ - **It cannot see a codebase it cannot fit in context** → a mechanically extracted
72
+ graph, queried instead of re-derived every session.
73
+ - **It fragments across sessions** → a spec, a plan and a task list that every session
74
+ and every swarm member builds against.
75
+ - **It builds what was asked instead of what was meant** → `/spec` marks unknowns
76
+ `[NEEDS CLARIFICATION]` and ends at a human.
77
+ - **It writes against the library it remembers, not the one installed** → a pinned,
78
+ per-role MCP server registry.
79
+
80
+ ## Install
81
+
82
+ macOS, Linux and Windows, from one artifact. Python 3.10+ is the whole requirement — no
83
+ dependencies, no compiler, no runtime. The package is
84
+ [`edify-agents-cli` on PyPI](https://pypi.org/project/edify-agents-cli/).
85
+
86
+ ```bash
87
+ pip install edify-agents-cli # or: pipx install / uv tool install edify-agents-cli
88
+ brew install edify-dev/tap/edify # macOS, if you prefer brew
89
+
90
+ cd your-repository
91
+ edify init # detect, extract the graph, install, write CLAUDE.md
92
+
93
+ edify init new ./service # or: a folder from zero, for every agent
94
+ ```
95
+
96
+ <p align="center">
97
+ <img src="assets/motion/init.webp" width="720" alt="edify init, in flat grey type, filling with blue ink">
98
+ </p>
99
+
100
+ `init` detects the stack, extracts the graph, scrapes conventions from your config
101
+ files, installs the skill entries your stack matches, seeds the server registry, writes
102
+ a fifteen-line `CLAUDE.md`, and records every file it put there. It asks one question —
103
+ which library entries to install, since they become trusted instruction — and with no
104
+ terminal (CI, a pipe, a model) it asks nothing.
105
+
106
+ <img src="assets/init.svg" width="100%" alt="edify init at a terminal: the wordmark with blue ink in it, then one line per step, ending on Next: /spec">
107
+
108
+ ```bash
109
+ edify init --yes # the whole matched set, no question
110
+ edify init --skills planner,verifier # exactly these
111
+ edify setup ./service --yes # a new folder, root pinned there
112
+ edify setup . --fresh --yes # reinstall the harness over this repository
113
+ ```
114
+
115
+ `edify init new` installs for every agent runtime at once, after eight skippable
116
+ questions whose answers go to `.edify/profile.md`:
117
+
118
+ | runtime | reads | commands land in |
119
+ |---|---|---|
120
+ | Claude Code | `CLAUDE.md` | `.claude/commands/`, `.claude/skills/`, `.claude/agents/` |
121
+ | OpenAI Codex | `AGENTS.md` | `.codex/prompts/` |
122
+ | Cursor | `AGENTS.md`, `.cursor/rules/edify.mdc` | `.cursor/commands/` |
123
+ | Gemini CLI | `GEMINI.md` | `.gemini/commands/` (TOML) |
124
+ | GitHub Copilot | `.github/copilot-instructions.md` | `.github/prompts/` |
125
+ | Windsurf | `.windsurf/rules/edify.md` | `.windsurf/workflows/` |
126
+
127
+ <table>
128
+ <tr>
129
+ <td><img src="assets/motion/rt-claude.webp" alt=".claude in ink"></td>
130
+ <td><img src="assets/motion/rt-codex.webp" alt=".codex in ink"></td>
131
+ <td><img src="assets/motion/rt-cursor.webp" alt=".cursor in ink"></td>
132
+ </tr>
133
+ <tr>
134
+ <td><img src="assets/motion/rt-gemini.webp" alt=".gemini in latex"></td>
135
+ <td><img src="assets/motion/rt-github.webp" alt=".github in blue latex"></td>
136
+ <td><img src="assets/motion/rt-windsurf.webp" alt=".windsurf in ink"></td>
137
+ </tr>
138
+ </table>
139
+
140
+ Per-platform detail, PATH fixes, updating, and uninstalling: [INSTALL.md](INSTALL.md).
141
+ It runs offline: only `edify upgrade` and `edify self update` open a socket.
142
+
143
+ ## The four commands
144
+
145
+ | command | what it does | ends at |
146
+ |---|---|---|
147
+ | `/spec` | a brief becomes a specification; unknowns marked `[NEEDS CLARIFICATION]` | a human |
148
+ | `/plan` | binding decisions, milestones, a partition for parallel work — only when needed | a human |
149
+ | `/tasks` | spec + plan + graph → leaseable tasks, **then builds them**: one agent or a swarm | a green suite |
150
+ | `/verify` | evidence on every surface — service, browser, device, live API — from a fresh session that wrote no code | a report anyone can read |
151
+
152
+ The build is the same loop at any scale:
153
+
154
+ ```bash
155
+ edify swarm claim # lease the next ready task nobody's work overlaps
156
+ edify swarm prompt T-7 # skill + task + decisions + graph slice + scoped servers
157
+ # … build it, run its done-check, commit on swarm/T-7 …
158
+ edify swarm done T-7 # the next claim may now unblock
159
+ ```
160
+
161
+ One agent runs it in a loop; a swarm runs it in N worktrees at once; people on
162
+ different machines add `--git`, and the remote's `edify-swarm` branch becomes the lock.
163
+ Leases are phase-gated and overlap-safe: a task whose files, or their dependents, touch
164
+ a live lease waits.
165
+
166
+ <table>
167
+ <tr>
168
+ <td width="50%"><img src="assets/motion/spec.webp" alt="spec, plan, tasks, read one after another"></td>
169
+ <td width="50%"><img src="assets/motion/build.webp" alt="A grid of failing red squares turning to passing blue, one by one"></td>
170
+ </tr>
171
+ <tr>
172
+ <td align="center"><sub><code>/spec</code> · <code>/plan</code> · <code>/tasks</code> — three reads, one after another</sub></td>
173
+ <td align="center"><sub>the <code>/tasks</code> run — red to green</sub></td>
174
+ </tr>
175
+ </table>
176
+
177
+ The command files live once in `.edify/commands/`, with thin pointers under each
178
+ runtime's command directory. Skills are mirrored the same way (`.claude/skills/<name>/SKILL.md`
179
+ points at `.edify/skills/<name>.md`); `edify skills sync` rewrites the mirror.
180
+ Verification comes first: phase 2 of every build turns each spec assertion into a
181
+ failing test before any core logic exists, so a cheap builder is hitting a target, not
182
+ judging its own work.
183
+
184
+ ## The graph
185
+
186
+ <p align="center">
187
+ <img src="assets/infinity.svg" width="480" alt="An infinity traced dot by dot — the animation the EDIFY CLI paints while it works">
188
+ </p>
189
+
190
+ Every file, module, exported symbol, route and table, with the edges between them, as
191
+ sorted checksummed text — a rebuild over unchanged source is byte-identical. **It is
192
+ never written by a model**: an uncovered language has no nodes and
193
+ `meta.languages_uncovered` says so. Python is parsed with a real syntax tree; the rest
194
+ by a line scanner, or by Universal Ctags with `--extractor ctags`.
195
+
196
+ ```bash
197
+ edify graph build # full extraction
198
+ edify graph where isExpired # → src/auth/clock.ts:12 symbol isExpired
199
+ edify graph dependents createSession # everything that breaks if it changes
200
+ edify graph overlap "a.ts b.ts" "c.ts"# can these two tasks run at once
201
+ ```
202
+
203
+ **Built for repositories no context window holds.** No agent reads a million-line
204
+ graph, so the graph summarises itself:
205
+
206
+ | command | answers |
207
+ |---|---|
208
+ | `edify graph map` | `MAP.md`: subsystems, hubs, and the bridges parallel work collides on |
209
+ | `edify graph context "<task>"` | the slice one task needs — files with line ranges, one hop of edges — under a token budget |
210
+ | `edify graph explain <x>` · `path <a> <b>` | every edge of one node, or how two connect |
211
+ | `edify graph update --changed` · `hook` | re-extract only what git says changed; git hooks keep it current |
212
+
213
+ Exit codes: 0 answered, 3 ambiguous (pin with `file::name`), 4 not found. Every other
214
+ command — `check`, `skills`, `mcp`, `governance`, `doctor`, `license`, `feedback`,
215
+ `self` — is in the [CLI reference](docs/cli.md).
216
+
217
+ ## What lands in your repository
218
+
219
+ ```
220
+ CLAUDE.md, AGENTS.md, … # ~15 lines each: what this repo is, the commands, the graph
221
+ .edify/
222
+ graph/ # nodes.tsv, edges.tsv, MAP.md — mechanical, regenerable
223
+ skills/ # the subset your stack matched, plus index.tsv
224
+ commands/ # the four command files
225
+ formats/ # the three format contracts and their worked examples
226
+ mcp.md # the server registry: pinned, scoped per role and phase
227
+ conventions.md # scraped from lint/format/tsconfig/CI — no model call
228
+ governance.tsv # every file above: origin, provenance, licence, hash
229
+ specs/<feature>/ # spec.md · plan.md · tasks.md · verification.md
230
+ ```
231
+
232
+ `edify governance verify` reports any installed file that changed since it landed.
233
+ Nothing blocks: `edify check` prints, and `--exit-code` is how you turn it into a gate
234
+ you own.
235
+
236
+ ## Editions, privacy, and what EDIFY does not do
237
+
238
+ <p align="center">
239
+ <img src="assets/motion/free.webp" width="720" alt="$0, filling with blue ink">
240
+ </p>
241
+
242
+ **The public edition is the whole product**: MIT-licensed, free, no account, any
243
+ repository size. Its one cap: the MCP registry scopes a single server into spawns. The
244
+ team edition, `edify-agents-teams`, is by licence and lifts it; a licence is a signed
245
+ token verified locally, issued on request. `edify license status` says which edition
246
+ this machine runs.
247
+
248
+ **There is no telemetry** — not anonymous, not aggregate, not opt-out. `edify feedback`
249
+ writes four answers to a file on your machine and prints how to send it, if you choose.
250
+
251
+ EDIFY does not wrap the agent runtime, host a model, run a server or daemon, or enforce
252
+ anything at runtime. It makes a competent model efficient; it does not make a weak one
253
+ competent. On a small solo project a frontier model in auto mode is already excellent —
254
+ the value starts where the codebase stops fitting in the context window.
255
+
256
+ ## Docs and development
257
+
258
+ [Quickstart](docs/quickstart.md) · [CLI](docs/cli.md) · [Graph](docs/graph.md) ·
259
+ [Agents](docs/agents.md) · [Verification](docs/verification.md) ·
260
+ [Privacy](docs/privacy.md) · [Troubleshooting](docs/troubleshooting.md) ·
261
+ [Changelog](CHANGELOG.md) · [Contributing](CONTRIBUTING.md)
262
+
263
+ ```bash
264
+ git clone https://github.com/EDIFY-agents/EDIFY && cd EDIFY
265
+ pip install -e ".[dev]"
266
+ pytest
267
+ ```
268
+
269
+ The suite checks that this repository satisfies what it asks of yours: every skill file
270
+ valid and under budget, every command file short, every worked example parsing against
271
+ its format.
272
+
273
+ ## Licence
274
+
275
+ See [LICENSE](LICENSE). Skill files carry their own provenance and licence in
276
+ frontmatter; entries marked `provenance: adapted` are governed by their upstream licence.