pythia-plsql 0.15.7__tar.gz → 0.15.8__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 (66) hide show
  1. {pythia_plsql-0.15.7/scripts/pythia_plsql.egg-info → pythia_plsql-0.15.8}/PKG-INFO +28 -18
  2. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/README.md +27 -17
  3. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/pyproject.toml +1 -1
  4. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/scripts/pythia.py +397 -128
  5. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8/scripts/pythia_plsql.egg-info}/PKG-INFO +28 -18
  6. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-apply/SKILL.md +12 -13
  7. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-autopilot/SKILL.md +4 -4
  8. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-deploy/SKILL.md +4 -3
  9. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-setup/SKILL.md +7 -4
  10. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-setup/reference/environment.md +6 -3
  11. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_mcp.py +24 -0
  12. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_phase3.py +1 -1
  13. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_phase6.py +300 -2
  14. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/LICENSE +0 -0
  15. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/comment-current.sql +0 -0
  16. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/compile-errors.sql +0 -0
  17. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/current-schema.sql +0 -0
  18. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/db-nls.sql +0 -0
  19. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/db-version.sql +0 -0
  20. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/dependencies.sql +0 -0
  21. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/explain-display.sql +0 -0
  22. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/impact.sql +0 -0
  23. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/invalid-objects.sql +0 -0
  24. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/name-occupants.sql +0 -0
  25. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/object-names.sql +0 -0
  26. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/object-source.sql +0 -0
  27. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/plscope-enabled.sql +0 -0
  28. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/plscope-statements.sql +0 -0
  29. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/plscope-usages.sql +0 -0
  30. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/schema-settings.sql +0 -0
  31. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/session-privileges.sql +0 -0
  32. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/similar-candidates.sql +0 -0
  33. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/source.sql +0 -0
  34. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/table-comments.sql +0 -0
  35. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/table-indexes.sql +0 -0
  36. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/table-stats.sql +0 -0
  37. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/queries/unindexed-fks.sql +0 -0
  38. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/scripts/pythia_plsql.egg-info/SOURCES.txt +0 -0
  39. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/scripts/pythia_plsql.egg-info/dependency_links.txt +0 -0
  40. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/scripts/pythia_plsql.egg-info/entry_points.txt +0 -0
  41. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/scripts/pythia_plsql.egg-info/requires.txt +0 -0
  42. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/scripts/pythia_plsql.egg-info/top_level.txt +0 -0
  43. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/setup.cfg +0 -0
  44. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-comment/SKILL.md +0 -0
  45. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-conventions/SKILL.md +0 -0
  46. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-explore/SKILL.md +0 -0
  47. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-explore/reference/data-dictionary.md +0 -0
  48. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-impact/SKILL.md +0 -0
  49. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-package/SKILL.md +0 -0
  50. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-package/reference/plsql-standards.md +0 -0
  51. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-review/SKILL.md +0 -0
  52. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-review/reference/antipatterns.md +0 -0
  53. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-skill-author/SKILL.md +0 -0
  54. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-spec/SKILL.md +0 -0
  55. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-table/SKILL.md +0 -0
  56. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-table/reference/ddl-standards.md +0 -0
  57. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-tune/SKILL.md +0 -0
  58. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-tune/reference/tuning-checklist.md +0 -0
  59. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-verify/SKILL.md +0 -0
  60. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-write/SKILL.md +0 -0
  61. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/pythia-write/reference/patterns.md +0 -0
  62. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/skills/using-pythia/SKILL.md +0 -0
  63. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_install.py +0 -0
  64. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_phase1.py +0 -0
  65. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_phase2.py +0 -0
  66. {pythia_plsql-0.15.7 → pythia_plsql-0.15.8}/tests/test_phase5.py +0 -0
@@ -1,6 +1,6 @@
1
1
  Metadata-Version: 2.4
2
2
  Name: pythia-plsql
3
- Version: 0.15.7
3
+ Version: 0.15.8
4
4
  Summary: PL/SQL development for AI agents on Oracle Database - expert data-dictionary queries, impact analysis, and a snapshot-verified write path with honest rollback.
5
5
  License: MIT
6
6
  Project-URL: Homepage, https://github.com/thaildhe172591/pythia
@@ -102,12 +102,15 @@ verify → report**. DDL self-commits in Oracle, so the snapshot is the only rea
102
102
  undo — it runs first and no flag disables it. A content-bound token guarantees
103
103
  what lands is byte-for-byte what was previewed, and a **developer-minted grant**
104
104
  guarantees a person approved it at all — through a door that is always the
105
- developer's. The `pythia_approve` tool (MCP, on Claude Code and Codex) shows
106
- pythia's own approval card as a select prompt, and Claude Code prompts for it
107
- in every permission mode but `dontAsk` — no allow rule can skip it. In chat:
108
- an `AskUserQuestion` whose text is the card (`pythia approve --card <token>`),
109
- whose **Approve** — written into the hook payload by Claude Code, never by
110
- the agent — mints the one-time grant. At a console: `pythia approve <token>`.
105
+ developer's, tried in order until one works. On Claude Code first: an
106
+ `AskUserQuestion` whose text is the card (`pythia approve --card <token>`) —
107
+ Claude Code's own dialog, shown in every client (terminal, desktop, web,
108
+ phone) and never answered by a hook or an allow rule; its **Approve**, chosen
109
+ in the dialog and traced through the session transcript to make sure the
110
+ agent did not write it, mints the one-time grant. Next, and first on Codex:
111
+ the `pythia_approve` tool (MCP), pythia's own card as a select prompt. Last, at
112
+ a console: `pythia approve <token>`. A door that fails says so and names the
113
+ next, and the agent moves on without asking.
111
114
  Without that grant `apply --confirm` refuses. A headless agent cannot `--yes`
112
115
  its own writes, cannot mint approval, and cannot loosen policy — each of those
113
116
  takes a human, answering or typing.
@@ -221,20 +224,25 @@ each agent needs you to finish by hand.
221
224
 
222
225
  Skills land in `.claude/skills` (or `~/.claude/skills` with `-g`).
223
226
  2. **Restart the Claude Code session** so the hooks and the MCP server load.
224
- 3. **Run `pythia doctor`.** It confirms the wiring and names every setting
225
- known to make `AskUserQuestion` return without showing the question — an
227
+ 3. **Run `pythia doctor`.** It confirms the wiring and names the settings
228
+ that once made `AskUserQuestion` return without showing the question — an
226
229
  allow rule or a skill's `allowed-tools` naming it, a PreToolUse hook with an
227
- empty or `*` matcher, `dontAsk` mode. That class of setting is why chat
228
- approval could fail outside `bypassPermissions`; the MCP door is immune
229
- to all but `dontAsk`.
230
+ empty or `*` matcher (in settings, or in a plugin that is installed and
231
+ enabled — the rest of a marketplace's plugins do not count). Current
232
+ Claude Code requires the dialog for `AskUserQuestion` whatever those say,
233
+ so they are warnings; `dontAsk` mode, which denies both chat doors, is the
234
+ one fail.
230
235
  4. *Optional:* copy the allow-list and `autoMode` from
231
236
  [`examples/claude-code-settings.example.json`](examples/claude-code-settings.example.json).
232
237
  Those are your permission preferences, so pythia does not set them for you
233
238
  ([why](GUIDE.md#11-optional-claude-code-permission-settings)).
234
239
 
235
- **Approving a change:** the agent previews with `pythia apply`, calls the
236
- `pythia_approve` tool (or asks an `AskUserQuestion` carrying the card), you
237
- click **Approve**, and the agent runs `apply --confirm` itself.
240
+ **Approving a change:** the agent previews with `pythia apply`, asks an
241
+ `AskUserQuestion` carrying the card, you click **Approve**, and the agent runs
242
+ `apply --confirm` itself. If that dialog's answer does not count (it closed
243
+ while you were away, or the hook could not trace it), the agent goes on
244
+ unasked to the `pythia_approve` tool, then to `pythia approve <token>` for your
245
+ terminal.
238
246
 
239
247
  **Working autonomously:** say so. The agent requests a lease
240
248
  (`pythia autopilot request`), you approve its card once, and it builds —
@@ -257,7 +265,9 @@ tables, comments, packages, releases — without stopping for each write.
257
265
  `pythia: connected`.
258
266
 
259
267
  Needs **Codex ≥ v0.120** for the in-chat approval (MCP elicitation); an older
260
- or headless Codex falls back to the console `pythia approve <token>`. Do
268
+ or headless Codex — or any host that declines the prompt without showing it,
269
+ which pythia tells apart by time — falls back to the console `pythia approve
270
+ <token>`, and the agent says so without being asked. Do
261
271
  **not** run Codex under `danger-full-access` — that sandbox auto-approves
262
272
  elicitations and bypasses your Approve.
263
273
 
@@ -313,8 +323,8 @@ works just as well.
313
323
  `pythia check` warns when the session runs with more power than the task needs.
314
324
 
315
325
  The developer approval that `apply` requires is minted per agent — Claude
316
- Code's `AskUserQuestion` hook, or Codex's `pythia_approve` MCP tool, or the
317
- console `pythia approve <token>` anywhere. How each is wired is in
326
+ Code's `AskUserQuestion` hook first, then the `pythia_approve` MCP tool (where
327
+ Codex starts), then the console `pythia approve <token>` anywhere. How each is wired is in
318
328
  [Setting up Claude Code and Codex](#setting-up-claude-code-and-codex) above.
319
329
 
320
330
  Per-group write policy, `.pythia/policy.json` (defaults shown):
@@ -88,12 +88,15 @@ verify → report**. DDL self-commits in Oracle, so the snapshot is the only rea
88
88
  undo — it runs first and no flag disables it. A content-bound token guarantees
89
89
  what lands is byte-for-byte what was previewed, and a **developer-minted grant**
90
90
  guarantees a person approved it at all — through a door that is always the
91
- developer's. The `pythia_approve` tool (MCP, on Claude Code and Codex) shows
92
- pythia's own approval card as a select prompt, and Claude Code prompts for it
93
- in every permission mode but `dontAsk` — no allow rule can skip it. In chat:
94
- an `AskUserQuestion` whose text is the card (`pythia approve --card <token>`),
95
- whose **Approve** — written into the hook payload by Claude Code, never by
96
- the agent — mints the one-time grant. At a console: `pythia approve <token>`.
91
+ developer's, tried in order until one works. On Claude Code first: an
92
+ `AskUserQuestion` whose text is the card (`pythia approve --card <token>`) —
93
+ Claude Code's own dialog, shown in every client (terminal, desktop, web,
94
+ phone) and never answered by a hook or an allow rule; its **Approve**, chosen
95
+ in the dialog and traced through the session transcript to make sure the
96
+ agent did not write it, mints the one-time grant. Next, and first on Codex:
97
+ the `pythia_approve` tool (MCP), pythia's own card as a select prompt. Last, at
98
+ a console: `pythia approve <token>`. A door that fails says so and names the
99
+ next, and the agent moves on without asking.
97
100
  Without that grant `apply --confirm` refuses. A headless agent cannot `--yes`
98
101
  its own writes, cannot mint approval, and cannot loosen policy — each of those
99
102
  takes a human, answering or typing.
@@ -207,20 +210,25 @@ each agent needs you to finish by hand.
207
210
 
208
211
  Skills land in `.claude/skills` (or `~/.claude/skills` with `-g`).
209
212
  2. **Restart the Claude Code session** so the hooks and the MCP server load.
210
- 3. **Run `pythia doctor`.** It confirms the wiring and names every setting
211
- known to make `AskUserQuestion` return without showing the question — an
213
+ 3. **Run `pythia doctor`.** It confirms the wiring and names the settings
214
+ that once made `AskUserQuestion` return without showing the question — an
212
215
  allow rule or a skill's `allowed-tools` naming it, a PreToolUse hook with an
213
- empty or `*` matcher, `dontAsk` mode. That class of setting is why chat
214
- approval could fail outside `bypassPermissions`; the MCP door is immune
215
- to all but `dontAsk`.
216
+ empty or `*` matcher (in settings, or in a plugin that is installed and
217
+ enabled — the rest of a marketplace's plugins do not count). Current
218
+ Claude Code requires the dialog for `AskUserQuestion` whatever those say,
219
+ so they are warnings; `dontAsk` mode, which denies both chat doors, is the
220
+ one fail.
216
221
  4. *Optional:* copy the allow-list and `autoMode` from
217
222
  [`examples/claude-code-settings.example.json`](examples/claude-code-settings.example.json).
218
223
  Those are your permission preferences, so pythia does not set them for you
219
224
  ([why](GUIDE.md#11-optional-claude-code-permission-settings)).
220
225
 
221
- **Approving a change:** the agent previews with `pythia apply`, calls the
222
- `pythia_approve` tool (or asks an `AskUserQuestion` carrying the card), you
223
- click **Approve**, and the agent runs `apply --confirm` itself.
226
+ **Approving a change:** the agent previews with `pythia apply`, asks an
227
+ `AskUserQuestion` carrying the card, you click **Approve**, and the agent runs
228
+ `apply --confirm` itself. If that dialog's answer does not count (it closed
229
+ while you were away, or the hook could not trace it), the agent goes on
230
+ unasked to the `pythia_approve` tool, then to `pythia approve <token>` for your
231
+ terminal.
224
232
 
225
233
  **Working autonomously:** say so. The agent requests a lease
226
234
  (`pythia autopilot request`), you approve its card once, and it builds —
@@ -243,7 +251,9 @@ tables, comments, packages, releases — without stopping for each write.
243
251
  `pythia: connected`.
244
252
 
245
253
  Needs **Codex ≥ v0.120** for the in-chat approval (MCP elicitation); an older
246
- or headless Codex falls back to the console `pythia approve <token>`. Do
254
+ or headless Codex — or any host that declines the prompt without showing it,
255
+ which pythia tells apart by time — falls back to the console `pythia approve
256
+ <token>`, and the agent says so without being asked. Do
247
257
  **not** run Codex under `danger-full-access` — that sandbox auto-approves
248
258
  elicitations and bypasses your Approve.
249
259
 
@@ -299,8 +309,8 @@ works just as well.
299
309
  `pythia check` warns when the session runs with more power than the task needs.
300
310
 
301
311
  The developer approval that `apply` requires is minted per agent — Claude
302
- Code's `AskUserQuestion` hook, or Codex's `pythia_approve` MCP tool, or the
303
- console `pythia approve <token>` anywhere. How each is wired is in
312
+ Code's `AskUserQuestion` hook first, then the `pythia_approve` MCP tool (where
313
+ Codex starts), then the console `pythia approve <token>` anywhere. How each is wired is in
304
314
  [Setting up Claude Code and Codex](#setting-up-claude-code-and-codex) above.
305
315
 
306
316
  Per-group write policy, `.pythia/policy.json` (defaults shown):
@@ -4,7 +4,7 @@ build-backend = "setuptools.build_meta"
4
4
 
5
5
  [project]
6
6
  name = "pythia-plsql"
7
- version = "0.15.7"
7
+ version = "0.15.8"
8
8
  description = "PL/SQL development for AI agents on Oracle Database - expert data-dictionary queries, impact analysis, and a snapshot-verified write path with honest rollback."
9
9
  readme = "README.md"
10
10
  license = { text = "MIT" }