diff --git a/.github/workflows/gh-aw-agent-suggestions.lock.yml b/.github/workflows/gh-aw-agent-suggestions.lock.yml
index 97031610..2ef325fc 100644
--- a/.github/workflows/gh-aw-agent-suggestions.lock.yml
+++ b/.github/workflows/gh-aw-agent-suggestions.lock.yml
@@ -316,7 +316,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-autonomy-atomicity-analyzer.lock.yml b/.github/workflows/gh-aw-autonomy-atomicity-analyzer.lock.yml
index b69d39be..92432ae8 100644
--- a/.github/workflows/gh-aw-autonomy-atomicity-analyzer.lock.yml
+++ b/.github/workflows/gh-aw-autonomy-atomicity-analyzer.lock.yml
@@ -316,7 +316,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-bug-hunter.lock.yml b/.github/workflows/gh-aw-bug-hunter.lock.yml
index c35c1d99..1974dc42 100644
--- a/.github/workflows/gh-aw-bug-hunter.lock.yml
+++ b/.github/workflows/gh-aw-bug-hunter.lock.yml
@@ -320,7 +320,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-code-simplifier.lock.yml b/.github/workflows/gh-aw-code-simplifier.lock.yml
index 72ef17e1..5641458b 100644
--- a/.github/workflows/gh-aw-code-simplifier.lock.yml
+++ b/.github/workflows/gh-aw-code-simplifier.lock.yml
@@ -299,7 +299,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-fragments/pick-three-keep-all.md b/.github/workflows/gh-aw-fragments/pick-three-keep-all.md
index 3f5c990b..ed9776c5 100644
--- a/.github/workflows/gh-aw-fragments/pick-three-keep-all.md
+++ b/.github/workflows/gh-aw-fragments/pick-three-keep-all.md
@@ -5,7 +5,7 @@ Parallelize your work using sub-agents. Spawn 3 sub-agents, each approaching the
 **How to spawn sub-agents:** Call `runSubagent` with `agentType: "general-purpose"` and `model: "${{ inputs.model }}"` (unless the workflow specifies a different agent type or model). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs:
 
 - The full task description and objective (restate it, don't summarize)
-- All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+- All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
 - Any relevant data the sub-agent needs to do its job (diffs, file contents, existing threads)
 - The quality criteria and output format you expect
 - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-fragments/pick-three-keep-one.md b/.github/workflows/gh-aw-fragments/pick-three-keep-one.md
index db1dd46d..b6a9a33d 100644
--- a/.github/workflows/gh-aw-fragments/pick-three-keep-one.md
+++ b/.github/workflows/gh-aw-fragments/pick-three-keep-one.md
@@ -5,7 +5,7 @@ Parallelize your initial investigation using sub-agents. Spawn 3 sub-agents, eac
 **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "${{ inputs.model }}"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
 
 - The full task description and objective (restate it, don't summarize)
-- All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+- All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
 - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
 - The quality gate criteria and output format you expect
 - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-information-architecture.lock.yml b/.github/workflows/gh-aw-information-architecture.lock.yml
index a3355023..fd93845f 100644
--- a/.github/workflows/gh-aw-information-architecture.lock.yml
+++ b/.github/workflows/gh-aw-information-architecture.lock.yml
@@ -316,7 +316,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-issue-triage.lock.yml b/.github/workflows/gh-aw-issue-triage.lock.yml
index 6418faa2..d7e0a30a 100644
--- a/.github/workflows/gh-aw-issue-triage.lock.yml
+++ b/.github/workflows/gh-aw-issue-triage.lock.yml
@@ -312,7 +312,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-performance-profiler.lock.yml b/.github/workflows/gh-aw-performance-profiler.lock.yml
index b331213a..97fd0f57 100644
--- a/.github/workflows/gh-aw-performance-profiler.lock.yml
+++ b/.github/workflows/gh-aw-performance-profiler.lock.yml
@@ -320,7 +320,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-product-manager-impersonator.lock.yml b/.github/workflows/gh-aw-product-manager-impersonator.lock.yml
index 66fa82e0..a2d0d9d2 100644
--- a/.github/workflows/gh-aw-product-manager-impersonator.lock.yml
+++ b/.github/workflows/gh-aw-product-manager-impersonator.lock.yml
@@ -328,7 +328,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-refactor-opportunist.lock.yml b/.github/workflows/gh-aw-refactor-opportunist.lock.yml
index 92c3f7e8..5118715a 100644
--- a/.github/workflows/gh-aw-refactor-opportunist.lock.yml
+++ b/.github/workflows/gh-aw-refactor-opportunist.lock.yml
@@ -319,7 +319,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-test-coverage-detector.lock.yml b/.github/workflows/gh-aw-test-coverage-detector.lock.yml
index 3278c144..c2428f42 100644
--- a/.github/workflows/gh-aw-test-coverage-detector.lock.yml
+++ b/.github/workflows/gh-aw-test-coverage-detector.lock.yml
@@ -316,7 +316,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/gh-aw-ux-design-patrol.lock.yml b/.github/workflows/gh-aw-ux-design-patrol.lock.yml
index f1cb063a..f9fa7018 100644
--- a/.github/workflows/gh-aw-ux-design-patrol.lock.yml
+++ b/.github/workflows/gh-aw-ux-design-patrol.lock.yml
@@ -325,7 +325,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/upgrade-check.lock.yml b/.github/workflows/upgrade-check.lock.yml
index 31cf2776..81ae978a 100644
--- a/.github/workflows/upgrade-check.lock.yml
+++ b/.github/workflows/upgrade-check.lock.yml
@@ -343,7 +343,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others
diff --git a/.github/workflows/workflow-patrol.lock.yml b/.github/workflows/workflow-patrol.lock.yml
index 5544016d..6f6579a5 100644
--- a/.github/workflows/workflow-patrol.lock.yml
+++ b/.github/workflows/workflow-patrol.lock.yml
@@ -343,7 +343,7 @@ jobs:
           **How to spawn sub-agents:** Call `runSubagent` with the `agentType` and `model` specified by the workflow instructions below (defaulting to `agentType: "general-purpose"` and `model: "__GH_AW_INPUTS_MODEL__"` if none are specified). Sub-agents cannot see your conversation history, the other sub-agents' results, or any context you have gathered so far. Each prompt must be **fully self-contained** — include everything the sub-agent needs to do its job:
           
           - The full task description and objective (restate it, don't summarize)
-          - All repository context, conventions, and constraints you've gathered (e.g., from `generate_agents_md`)
+          - All repository context, conventions, and constraints you've gathered (e.g., from `/tmp/agents.md`)
           - Any relevant findings from your initial exploration (file paths, git log output, code snippets)
           - The quality gate criteria and output format you expect
           - The specific angle that distinguishes this sub-agent from the others