recall-os 0.2.1 → 0.3.0

package/examples/generated-flutter/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-flutter/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-flutter/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-generic/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-generic/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-generic/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-ios-swift/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-ios-swift/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-ios-swift/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-kotlin-android/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-kotlin-android/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-kotlin-android/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-laravel-api/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-laravel-api/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-laravel-api/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-laravel-react/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-laravel-react/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-laravel-react/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-laravel-vue/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-laravel-vue/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-laravel-vue/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-nextjs/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-nextjs/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-nextjs/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/examples/generated-python-fastapi/AGENTS.md

@@ -13,3 +13,16 @@ Required reading:
 - `docs/60-engineering/ENGINEERING_STANDARDS.md`
 
 Repository rules override model preferences. If instructions conflict, stop and report the conflict.
+
+## Changing an accepted decision
+
+Before changing anything an accepted ADR governs (framework, database, auth, API shape, and
+similar):
+
+1. Check `docs/adrs/` for an accepted ADR that covers it.
+2. If your change contradicts one, stop and confirm with a human first — do not silently change the
+   code and leave the ADR saying the opposite.
+3. Record the change as a new decision with `recall adr supersede <old> <new-title>`. That
+   supersedes the old ADR instead of overwriting history, so the reasoning stays auditable.
+
+Repository memory is only trustworthy if decisions change through this trail, not silently.

package/examples/generated-python-fastapi/docs/60-engineering/AI_AGENT_RULES.md

@@ -4,3 +4,12 @@ AI agents must follow repository memory over model preference.
 
 If a request conflicts with accepted repository memory or engineering standards, stop and report the
 conflict.
+
+## Changing an accepted decision
+
+When work would change something an accepted ADR governs:
+
+1. Find the accepted ADR in `docs/adrs/` that covers it.
+2. If the change contradicts it, stop and confirm with a human before changing the code.
+3. Record the new decision with `recall adr supersede <old> <new-title>` so the old ADR is marked
+   superseded and the reasoning is preserved, instead of silently editing or contradicting it.

package/examples/generated-python-fastapi/docs/ai/RECALL_COMMANDS.md

@@ -108,6 +108,17 @@ Options:
 - `--dry-run`: show planned writes without writing files.
 - `--force`: overwrite existing files explicitly.
 
+### `recall adr supersede <old> <new-title>`
+
+Record a changed decision. Marks an accepted ADR as `Accepted — superseded by ADR-####` and creates
+a new accepted ADR that declares what it supersedes, so the reasoning trail stays auditable instead
+of being overwritten. Doctor then warns about any memory still referencing the superseded decision.
+
+Options:
+
+- `--dry-run`: show planned writes without writing files.
+- `--force`: overwrite existing files explicitly.
+
 ### `recall module create <name>`
 
 Create module memory docs under the configured modules directory.
@@ -124,7 +135,8 @@ engineering evidence is present, and whether memory references decisions that ex
 accepted.
 
 Doctor also runs deterministic drift checks: feature or module memory that references a missing ADR
-is an error, and memory that references a not-yet-accepted ADR is a warning.
+is an error, memory that references a not-yet-accepted ADR is a warning, and memory that still
+references a superseded decision is a warning.
 
 Exit codes:
 

package/package.json

@@ -1,12 +1,12 @@
 {
   "name": "recall-os",
-  "version": "0.2.1",
+  "version": "0.3.0",
   "description": "Architecture-neutral repository memory for AI-assisted software teams.",
   "packageManager": "pnpm@10.12.3",
   "type": "module",
   "license": "MIT",
   "author": "Karthick Ramachandran",
-  "homepage": "https://github.com/Karthick-Ramachandran/recall-os#readme",
+  "homepage": "https://recall-os.pages.dev",
   "repository": {
     "type": "git",
     "url": "git+https://github.com/Karthick-Ramachandran/recall-os.git"