Skip to content

t1455: add adaptive pulse queue-governor and strict worker counting #4218

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
marcusquinn merged 1 commit into from
Mar 12, 2026
Merged

t1455: add adaptive pulse queue-governor and strict worker counting #4218

Show file tree
Hide file tree
Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
23 changes: 23 additions & 0 deletions .agents/scripts/commands/pulse.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -50,10 +50,17 @@ AVAILABLE=$((MAX_WORKERS - WORKER_COUNT))
# shows PRODUCT_MIN and TOOLING_MAX. Read these values:
PRODUCT_MIN=$(grep '^PRODUCT_MIN=' ~/.aidevops/logs/pulse-priority-allocations 2>/dev/null | cut -d= -f2 || echo 0)
TOOLING_MAX=$(grep '^TOOLING_MAX=' ~/.aidevops/logs/pulse-priority-allocations 2>/dev/null | cut -d= -f2 || echo "$MAX_WORKERS")

# Adaptive queue governor (t1455) from pre-fetched state
PULSE_QUEUE_MODE=$(grep '^PULSE_QUEUE_MODE=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo "balanced")
PR_REMEDIATION_FOCUS_PCT=$(grep '^PR_REMEDIATION_FOCUS_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo 50)
NEW_ISSUE_DISPATCH_PCT=$(grep '^NEW_ISSUE_DISPATCH_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo 50)
Comment on lines +55 to +57
Copy link
Contributor

@coderabbitai coderabbitai bot Mar 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟡 Minor

Multi-value grep returns could break variable comparisons.

If pulse-state.txt contains multiple PULSE_QUEUE_MODE= lines (e.g., from repeated governor calls or file corruption), grep returns all matching lines. The variable then contains newlines, causing string comparisons like [[ "$PULSE_QUEUE_MODE" == "pr-heavy" ]] to fail unexpectedly.

🛡️ Suggested fix using tail -1 to get the last (most recent) value 
-PULSE_QUEUE_MODE=$(grep '^PULSE_QUEUE_MODE=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo "balanced")
-PR_REMEDIATION_FOCUS_PCT=$(grep '^PR_REMEDIATION_FOCUS_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo 50)
-NEW_ISSUE_DISPATCH_PCT=$(grep '^NEW_ISSUE_DISPATCH_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo 50)
+PULSE_QUEUE_MODE=$(grep '^PULSE_QUEUE_MODE=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | tail -1 | cut -d= -f2 || echo "balanced")
+PR_REMEDIATION_FOCUS_PCT=$(grep '^PR_REMEDIATION_FOCUS_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | tail -1 | cut -d= -f2 || echo 50)
+NEW_ISSUE_DISPATCH_PCT=$(grep '^NEW_ISSUE_DISPATCH_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | tail -1 | cut -d= -f2 || echo 50)
📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change
PULSE_QUEUE_MODE=$(grep '^PULSE_QUEUE_MODE=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo "balanced")
PR_REMEDIATION_FOCUS_PCT=$(grep '^PR_REMEDIATION_FOCUS_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo 50)
NEW_ISSUE_DISPATCH_PCT=$(grep '^NEW_ISSUE_DISPATCH_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | cut -d= -f2 || echo 50)
PULSE_QUEUE_MODE=$(grep '^PULSE_QUEUE_MODE=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | tail -1 | cut -d= -f2 || echo "balanced")
PR_REMEDIATION_FOCUS_PCT=$(grep '^PR_REMEDIATION_FOCUS_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | tail -1 | cut -d= -f2 || echo 50)
NEW_ISSUE_DISPATCH_PCT=$(grep '^NEW_ISSUE_DISPATCH_PCT=' ~/.aidevops/logs/pulse-state.txt 2>/dev/null | tail -1 | cut -d= -f2 || echo 50)
🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In @.agents/scripts/commands/pulse.md around lines 55 - 57, The current
assignments for PULSE_QUEUE_MODE, PR_REMEDIATION_FOCUS_PCT, and
NEW_ISSUE_DISPATCH_PCT can capture multiple matching lines from pulse-state.txt
(introducing newlines and breaking comparisons); update each command that uses
grep to select a single value (e.g., pipe the grep output to tail -1 to pick the
most recent match) before cutting the value so the variables contain a single
scalar string suitable for comparisons like [[ "$PULSE_QUEUE_MODE" == "pr-heavy"
]].

```

If `AVAILABLE <= 0`: you can still merge ready PRs, but don't dispatch new workers.

If `PULSE_QUEUE_MODE` is `pr-heavy` or `merge-heavy`, spend most dispatch capacity on existing PR advancement (merge-ready first, then failing checks/changes requested) and limit new issue starts to the adaptive budget (`NEW_ISSUE_DISPATCH_PCT`).

### Priority-class enforcement (t1423)

Worker slots are partitioned between **product** repos (`"priority": "product"` in repos.json) and **tooling** repos (`"priority": "tooling"`). Product repos get a guaranteed minimum share (default 60%) to prevent tooling hygiene from starving user-facing work.
Expand Down Expand Up @@ -83,6 +90,14 @@ Scan the pre-fetched state above. Act immediately on each item — don't build a

### PRs — merge, fix, or flag

**Adaptive mode rule (t1455):**

- `merge-heavy`: complete merge-ready PRs first, then PR fix workers, then only minimal new issue dispatch.
- `pr-heavy`: prioritize PR repair/merge throughput over opening fresh issue work.
- `balanced`: normal ordering.

Treat this as a live queue-pressure signal, not a static threshold.

**External contributor gate (MANDATORY):** Before merging ANY PR, check if the author is a repo collaborator. The permission check must **fail closed** — if the API call itself fails, do NOT auto-merge and do NOT assume the author is external.

```bash
Expand Down Expand Up @@ -533,6 +548,14 @@ If decomposition succeeds (`SUBTASK_COUNT >= 2`):

For each dispatchable issue (intelligence-first):

When `PULSE_QUEUE_MODE` is `pr-heavy` or `merge-heavy`, limit issue dispatches to the current cycle budget:

```bash
ISSUE_DISPATCH_BUDGET=$(((AVAILABLE * NEW_ISSUE_DISPATCH_PCT) / 100))
```

If budget is exhausted, stop opening new issue workers and continue PR advancement work.

1. Skip if a worker is already running for it locally (check `ps` output for the issue number)
2. Skip if an open PR already exists for it (check PR list)
3. Treat labels as hints, not gates. `status:queued`, `status:in-progress`, and `status:in-review` suggest active work, but verify with evidence (active worker, recent PR updates, recent commits) before skipping.
Expand Down
106 changes: 105 additions & 1 deletion .agents/scripts/pulse-wrapper.sh
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -129,6 +129,7 @@ PULSE_MODEL="${PULSE_MODEL:-}"
HEADLESS_RUNTIME_HELPER="${HEADLESS_RUNTIME_HELPER:-${SCRIPT_DIR}/headless-runtime-helper.sh}"
REPOS_JSON="${REPOS_JSON:-${HOME}/.config/aidevops/repos.json}"
STATE_FILE="${HOME}/.aidevops/logs/pulse-state.txt"
QUEUE_METRICS_FILE="${HOME}/.aidevops/logs/pulse-queue-metrics"

if [[ ! -x "$HEADLESS_RUNTIME_HELPER" ]]; then
printf '[pulse-wrapper] ERROR: headless runtime helper is missing or not executable: %s (SCRIPT_DIR=%s)\n' "$HEADLESS_RUNTIME_HELPER" "$SCRIPT_DIR" >&2
Expand Down Expand Up @@ -347,6 +348,9 @@ prefetch_state() {
# Append priority-class worker allocations (t1423)
_append_priority_allocations >>"$STATE_FILE"

# Append adaptive queue-governor guidance (t1455)
append_adaptive_queue_governor

# Export PULSE_SCOPE_REPOS — comma-separated list of repo slugs that
# workers are allowed to create PRs/branches on (t1405, GH#2928).
# Workers CAN file issues on any repo (cross-repo self-improvement),
Expand Down Expand Up @@ -1678,11 +1682,111 @@ prefetch_contribution_watch() {
#######################################
count_active_workers() {
local count
count=$(ps axo command | grep '[/]full-loop' | grep -c '[.]opencode') || count=0
count=$(ps axo command | grep '\.opencode run' | grep '/full-loop Implement issue #' | grep -v '/pulse' | grep -c -v grep) || count=0
Copy link

@gemini-code-assist gemini-code-assist bot Mar 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The use of grep -c -v grep is a bit unconventional for counting processes while excluding the grep command itself. While it works by counting lines that don't contain "grep", a more common and arguably clearer idiom is to use grep -v grep | wc -l or to use the bracket trick in one of the patterns (e.g., grep '[.]opencode') to prevent grep from matching its own process. This would improve readability and maintainability for other developers.

Suggested change
count=$(ps axo command | grep '\.opencode run' | grep '/full-loop Implement issue #' | grep -v '/pulse' | grep -c -v grep) || count=0
count=$(ps axo command | grep '\.opencode run' | grep '/full-loop Implement issue #' | grep -v '/pulse' | grep -v grep | wc -l) || count=0

echo "$count"
return 0
}
Comment on lines 1683 to 1688
Copy link
Contributor

@coderabbitai coderabbitai bot Mar 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

⚠️ Potential issue | 🟡 Minor

🧩 Analysis chain

🏁 Script executed:

#!/bin/bash
# Check what worker dispatch patterns exist in the codebase
echo "=== Dispatch patterns in pulse.md ==="
rg --no-heading -n '/full-loop' .agents/scripts/commands/pulse.md | head -30

echo ""
echo "=== Mission worker dispatch patterns ==="
rg --no-heading -n 'headless-runtime-helper.sh run' .agents/scripts/commands/pulse.md | head -20

echo ""
echo "=== count_active_workers implementations ==="
rg --no-heading -A5 'count_active_workers\(\)' .agents/scripts/

Repository: marcusquinn/aidevops

Length of output: 3918

🏁 Script executed:

#!/bin/bash
# Check how count_active_workers is used in pulse-wrapper.sh
echo "=== Usage of count_active_workers in pulse-wrapper.sh ==="
rg --no-heading -B5 -A5 'count_active_workers\(\)' .agents/scripts/pulse-wrapper.sh | head -50

echo ""
echo "=== MAX_WORKERS and dispatch logic in pulse-wrapper.sh ==="
rg --no-heading -n 'MAX_WORKERS\|WORKER_COUNT\|count_active_workers' .agents/scripts/pulse-wrapper.sh | head -30

echo ""
echo "=== Check if pulse-wrapper.sh dispatches mission workers ==="
rg --no-heading -n 'mission\|Mission' .agents/scripts/pulse-wrapper.sh | head -20

Repository: marcusquinn/aidevops

Length of output: 584

🏁 Script executed:

#!/bin/bash
# Search for where count_active_workers is actually called
echo "=== Search for count_active_workers function calls ==="
rg --no-heading -n 'count_active_workers' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check what pulse-wrapper.sh dispatches ==="
rg --no-heading -n 'headless-runtime-helper\|\.opencode run' .agents/scripts/pulse-wrapper.sh | head -20

echo ""
echo "=== Get overall structure/purpose of pulse-wrapper.sh ==="
head -100 .agents/scripts/pulse-wrapper.sh | grep -E '^\#|^[A-Za-z_]+\(\)|script'

echo ""
echo "=== Check if count_active_workers is called anywhere ==="
rg --no-heading -n '\$count_active_workers\|count_active_workers' .agents/scripts/ --type sh

Repository: marcusquinn/aidevops

Length of output: 3237

🏁 Script executed:

#!/bin/bash
# Check context around line 1954 where count_active_workers is called
echo "=== Context around line 1954 (active_workers=$(count_active_workers)) ==="
sed -n '1945,1975p' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check for any logging or usage of active_workers variable ==="
rg --no-heading -B2 -A2 'active_workers=' .agents/scripts/pulse-wrapper.sh

Repository: marcusquinn/aidevops

Length of output: 2135

🏁 Script executed:

#!/bin/bash
# Check if mission/PR workers are dispatched from other contexts
echo "=== Check all worker dispatch points ==="
rg --no-heading -n 'headless-runtime-helper.sh run' .agents/scripts/*.sh | grep -v pulse.md

echo ""
echo "=== Check if there's queue enforcement for all worker types ==="
rg --no-heading -B5 -A5 'max_workers\|MAX_WORKERS' .agents/scripts/pulse-wrapper.sh | head -40

echo ""
echo "=== Check purpose/scope comments in pulse-wrapper.sh header ==="
head -50 .agents/scripts/pulse-wrapper.sh | tail -20

Repository: marcusquinn/aidevops

Length of output: 1370

🏁 Script executed:

#!/bin/bash
# Check if pulse-wrapper.sh dispatches any workers or just manages the pulse
echo "=== Search for dispatch/worker dispatch in pulse-wrapper.sh ==="
rg --no-heading -n 'dispatch\|queue\|Implement issue' .agents/scripts/pulse-wrapper.sh | head -20

echo ""
echo "=== Check if count_runnable_candidates matches same pattern ==="
rg --no-heading -B2 -A8 'count_runnable_candidates\(\)' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check what queued issues look like ==="
rg --no-heading -B2 -A8 'count_queued_without_worker\(\)' .agents/scripts/pulse-wrapper.sh

Repository: marcusquinn/aidevops

Length of output: 785

🏁 Script executed:

#!/bin/bash
# Check where mission and PR fix workers are actually dispatched from
echo "=== Check what pulse-wrapper.sh actually calls/dispatches ==="
rg --no-heading -n 'function\|^[a-z_]*() {' .agents/scripts/pulse-wrapper.sh | head -30

echo ""
echo "=== Look for dispatch_* or similar functions in pulse-wrapper ==="
rg --no-heading -B3 -A10 'dispatch\|enqueue' .agents/scripts/pulse-wrapper.sh | head -50

echo ""
echo "=== Check if there's a separate handler for mission/PR workers ==="
ls -la .agents/scripts/*.sh | grep -E 'mission|pr|queue|dispatch'

echo ""
echo "=== Look at mission-dashboard-helper to see if it has its own enforcement ==="
rg --no-heading -n 'max_workers\|count_active_workers' .agents/scripts/mission-dashboard-helper.sh

Repository: marcusquinn/aidevops

Length of output: 2376

🏁 Script executed:

#!/bin/bash
# Check function definitions without problematic regex
echo "=== Main functions in pulse-wrapper.sh ==="
rg --no-heading -n '^[a-z_]+\(\)' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check if mission-dashboard-helper has worker enforcement ==="
rg --no-heading -B5 -A10 'count_active_workers\|max_workers' .agents/scripts/mission-dashboard-helper.sh | head -50

echo ""
echo "=== Check cron-dispatch.sh for mission/PR dispatch logic ==="
rg --no-heading -n 'mission\|PR\|Address unresolved' .agents/scripts/cron-dispatch.sh | head -20

Repository: marcusquinn/aidevops

Length of output: 1220

🏁 Script executed:

#!/bin/bash
# Check what prefetch_active_workers does
echo "=== prefetch_active_workers function ==="
sed -n '880,940p' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check append_adaptive_queue_governor ==="
sed -n '1697,1750p' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check if pulse-wrapper.sh has any worker TYPE distinction ==="
rg --no-heading -n 'issue\|mission\|PR.*fix\|Address unresolved' .agents/scripts/pulse-wrapper.sh | head -20

Repository: marcusquinn/aidevops

Length of output: 4408

🏁 Script executed:

#!/bin/bash
# Check if enforce_utilization_invariants actually dispatches workers
echo "=== Full enforce_utilization_invariants function ==="
sed -n '1942,2000p' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check what calls enforce_utilization_invariants ==="
rg --no-heading -n 'enforce_utilization_invariants' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check if pulse-wrapper dispatches PR fix or mission workers ==="
rg --no-heading -B3 -A3 'PR\|mission\|Address unresolved\|Implement.*task' .agents/scripts/pulse-wrapper.sh | head -40

Repository: marcusquinn/aidevops

Length of output: 2780

🏁 Script executed:

#!/bin/bash
# Clarify: check if the narrow pattern is intentional by seeing how prefetch_active_workers is used
echo "=== Where is prefetch_active_workers output used? ==="
rg --no-heading -B5 -A5 'prefetch_active_workers' .agents/scripts/pulse-wrapper.sh | grep -v '^--$'

echo ""
echo "=== Check if mission workers or PR fix workers are explicitly SEPARATE from issue workers ==="
rg --no-heading -n 'run_pulse\|count_runnable_candidates\|count_queued_without_worker' .agents/scripts/pulse-wrapper.sh | head -5

echo ""
echo "=== Look at what run_pulse actually does ==="
sed -n '1289,1350p' .agents/scripts/pulse-wrapper.sh

echo ""
echo "=== Check if there's documentation about worker types in pulse.md ==="
head -100 .agents/scripts/commands/pulse.md | tail -50

Repository: marcusquinn/aidevops

Length of output: 8189

Backfill capacity check should account for all active worker types, not just issue workers.

The pattern /full-loop Implement issue # counts only issue workers. During backfill, enforce_utilization_invariants() exits when active_workers >= max_workers, but if PR-fix or mission workers are concurrently active, this check undercounts total capacity usage. This could cause the pulse agent to dispatch additional issue workers even when the system has already hit max_workers.

While prefetch_active_workers() correctly shows all /full-loop workers to the pulse agent, the backfill loop's separate capacity gate should match that scope. Consider broadening the pattern to grep '/full-loop' (excluding /pulse to avoid self-counting) or explicitly documenting that backfill is issue-only and relies on the pulse agent's prompt context to enforce cross-worker-type limits.

🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In @.agents/scripts/pulse-wrapper.sh around lines 1683 - 1688, The backfill
capacity check in count_active_workers() is only matching issue workers and
undercounts total active workers; update count_active_workers() to broaden the
grep pattern (e.g., match '/full-loop' while still excluding '/pulse') so it
reports all active full-loop workers, ensuring enforce_utilization_invariants()
uses the correct total when comparing active_workers to max_workers; locate
function count_active_workers and change the grep filter accordingly (or
alternatively document/rename it if you intend it to be issue-only).


#######################################
# Append adaptive queue-governor guidance to pre-fetched state
#
# Uses observed queue totals and trend vs previous cycle to derive an
# adaptive PR-vs-issue dispatch focus. This avoids static per-repo
# thresholds and shifts effort toward PR burn-down when PR backlog grows.
#######################################
append_adaptive_queue_governor() {
if [[ ! -f "$STATE_FILE" ]]; then
return 0
fi

local total_prs total_issues ready_prs failing_prs
total_prs=$(awk '/^### Open PRs \([0-9]+\)/ { line=$0; gsub(/[^0-9]/, "", line); sum+=line } END { print sum+0 }' "$STATE_FILE")
total_issues=$(awk '/^### Open Issues \([0-9]+\)/ { line=$0; gsub(/[^0-9]/, "", line); sum+=line } END { print sum+0 }' "$STATE_FILE")
ready_prs=$(rg -c '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
failing_prs=$(rg -c '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
Comment on lines +1705 to +1706
Copy link

@gemini-code-assist gemini-code-assist bot Mar 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

medium

The script relies on rg (ripgrep) to calculate ready_prs and failing_prs. While it safely defaults to 0 if rg is not found, this can lead to the adaptive governor making suboptimal decisions without any warning to the user. It would be beneficial to explicitly check for rg and log a warning if it's missing, informing the user that the feature's effectiveness is reduced. This aligns with the principle of managing external dependencies in portable shell scripts, ensuring users are aware when non-built-in tools are not available, even if a fallback is in place.

Suggested change
ready_prs=$(rg -c '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
failing_prs=$(rg -c '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
if ! command -v rg &>/dev/null; then
echo "[pulse-wrapper] WARNING: ripgrep (rg) not found. Adaptive queue governor metrics for ready/failing PRs will be disabled." >>"$LOGFILE"
ready_prs=0
failing_prs=0
else
ready_prs=$(rg -c '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
failing_prs=$(rg -c '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
fi
References
  1. For shell scripts that need to be portable, prefer built-in tools like grep over external dependencies like jq for simple tasks, to avoid installation requirements on diverse systems.

Comment on lines +1705 to +1706
Copy link
Contributor

@coderabbitai coderabbitai bot Mar 12, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

🛠️ Refactor suggestion | 🟠 Major

Use grep instead of rg for consistency and portability.

The script uses grep throughout, but this function introduces a rg (ripgrep) dependency. While the || echo "0" fallback prevents failures when rg is unavailable, it silently returns incorrect counts rather than using the actual data.

♻️ Suggested fix using grep 
-	ready_prs=$(rg -c '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
-	failing_prs=$(rg -c '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
+	ready_prs=$(grep -cE '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
+	failing_prs=$(grep -cE '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
📝 Committable suggestion

‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.

Suggested change
ready_prs=$(rg -c '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
failing_prs=$(rg -c '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
ready_prs=$(grep -cE '\[checks: PASS\].*\[review: APPROVED\]' "$STATE_FILE" 2>/dev/null || echo "0")
failing_prs=$(grep -cE '\[checks: FAIL\]|\[review: CHANGES_REQUESTED\]' "$STATE_FILE" 2>/dev/null || echo "0")
🤖 Prompt for AI Agents 
Verify each finding against the current code and only fix it if needed.

In @.agents/scripts/pulse-wrapper.sh around lines 1705 - 1706, Replace the
ripgrep calls that set ready_prs and failing_prs with portable grep invocations:
use grep -E -c with the same regex patterns against "$STATE_FILE" and keep the
2>/dev/null || echo "0" fallback so missing files still yield "0"; update the
lines that assign ready_prs and failing_prs (referencing those variable names
and STATE_FILE) to use grep -E and the original patterns '\[checks:
PASS\].*\[review: APPROVED\]' and '\[checks: FAIL\]|\[review:
CHANGES_REQUESTED\]'.


[[ "$total_prs" =~ ^[0-9]+$ ]] || total_prs=0
[[ "$total_issues" =~ ^[0-9]+$ ]] || total_issues=0
[[ "$ready_prs" =~ ^[0-9]+$ ]] || ready_prs=0
[[ "$failing_prs" =~ ^[0-9]+$ ]] || failing_prs=0

local prev_total_prs=0 prev_total_issues=0 prev_ready_prs=0 prev_failing_prs=0
if [[ -f "$QUEUE_METRICS_FILE" ]]; then
while IFS='=' read -r key value; do
case "$key" in
prev_total_prs) prev_total_prs="$value" ;;
prev_total_issues) prev_total_issues="$value" ;;
prev_ready_prs) prev_ready_prs="$value" ;;
prev_failing_prs) prev_failing_prs="$value" ;;
esac
done <"$QUEUE_METRICS_FILE"
fi

[[ "$prev_total_prs" =~ ^-?[0-9]+$ ]] || prev_total_prs=0
[[ "$prev_total_issues" =~ ^-?[0-9]+$ ]] || prev_total_issues=0
[[ "$prev_ready_prs" =~ ^-?[0-9]+$ ]] || prev_ready_prs=0
[[ "$prev_failing_prs" =~ ^-?[0-9]+$ ]] || prev_failing_prs=0

local pr_delta issue_delta ready_delta failing_delta
pr_delta=$((total_prs - prev_total_prs))
issue_delta=$((total_issues - prev_total_issues))
ready_delta=$((ready_prs - prev_ready_prs))
failing_delta=$((failing_prs - prev_failing_prs))

local denominator pr_share_pct growth_bias pr_focus_pct new_issue_pct
denominator=$((total_prs + total_issues))
if [[ "$denominator" -lt 1 ]]; then
denominator=1
fi
pr_share_pct=$(((total_prs * 100) / denominator))
growth_bias=0
if [[ "$pr_delta" -gt 0 ]]; then
growth_bias=10
elif [[ "$pr_delta" -lt 0 ]]; then
growth_bias=-5
fi
pr_focus_pct=$((35 + (pr_share_pct / 2) + growth_bias))
if [[ "$pr_focus_pct" -lt 35 ]]; then
pr_focus_pct=35
elif [[ "$pr_focus_pct" -gt 85 ]]; then
pr_focus_pct=85
fi
new_issue_pct=$((100 - pr_focus_pct))

local queue_mode
queue_mode="balanced"
if [[ "$ready_prs" -gt 0 && "$pr_delta" -ge 0 ]]; then
queue_mode="merge-heavy"
elif [[ "$pr_focus_pct" -ge 60 ]]; then
queue_mode="pr-heavy"
fi

cat >"$QUEUE_METRICS_FILE" <<EOF
prev_total_prs=${total_prs}
prev_total_issues=${total_issues}
prev_ready_prs=${ready_prs}
prev_failing_prs=${failing_prs}
EOF

{
echo ""
echo "## Adaptive Queue Governor"
echo "- Queue totals: PRs=${total_prs} (delta ${pr_delta}), issues=${total_issues} (delta ${issue_delta})"
echo "- PR execution pressure: ready=${ready_prs} (delta ${ready_delta}), failing_or_changes_requested=${failing_prs} (delta ${failing_delta})"
echo "- Adaptive mode this cycle: ${queue_mode}"
echo "- Recommended dispatch focus: PR remediation ${pr_focus_pct}% / new issue dispatch ${new_issue_pct}%"
echo ""
echo "PULSE_QUEUE_MODE=${queue_mode}"
echo "PR_REMEDIATION_FOCUS_PCT=${pr_focus_pct}"
echo "NEW_ISSUE_DISPATCH_PCT=${new_issue_pct}"
echo ""
echo "When PR backlog is rising, prioritize merge-ready and failing-check PR advancement before new issue starts."
} >>"$STATE_FILE"

echo "[pulse-wrapper] Adaptive queue governor: mode=${queue_mode} prs=${total_prs} issues=${total_issues} pr_focus=${pr_focus_pct}%" >>"$LOGFILE"
return 0
}

#######################################
# Get current max workers from pulse-max-workers file
# Returns: numeric value via stdout (defaults to 1)
Expand Down
Loading