Skip to content

Add debugging #53

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
derekmisler merged 1 commit into from
Feb 26, 2026
Merged

Add debugging #53

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
3 changes: 1 addition & 2 deletions README.md
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -104,7 +104,6 @@ See the [full PR Review documentation](review-pr/README.md) for more details.
yolo: false # Require manual approval
timeout: 600 # 10 minute timeout
debug: true # Enable debug logging
quiet: false # Show verbose tool calls (default: true)
working-directory: ./src
extra-args: "--verbose"
add-prompt-files: "AGENTS.md,CLAUDE.md" # Append these files to the prompt
Expand Down Expand Up @@ -159,7 +158,6 @@ See the [full PR Review documentation](review-pr/README.md) for more details.
| `debug` | Enable debug mode with verbose logging (`true`/`false`) | No | `false` |
| `working-directory` | Working directory to run the agent in | No | `.` |
| `yolo` | Auto-approve all prompts (`true`/`false`) | No | `true` |
| `quiet` | Suppress verbose tool call output (`true`/`false`). Set to `false` for debugging. | No | `true` |
| `max-retries` | Maximum number of retries on failure (0 = no retries) | No | `2` |
| `retry-delay` | Base delay in seconds between retries (doubles each attempt) | No | `5` |
| `extra-args` | Additional arguments to pass to `cagent run` | No | - |
Expand Down Expand Up @@ -197,6 +195,7 @@ add-prompt-files: "AGENTS.md" # Found via hierarchy search
| `cagent-version` | Version of cagent that was used |
| `mcp-gateway-installed` | Whether mcp-gateway was installed (`true`/`false`) |
| `execution-time` | Agent execution time in seconds |
| `verbose-log-file` | Path to the full verbose agent log (includes tool calls) |
| `secrets-detected` | Whether secrets were detected in output |
| `prompt-suspicious` | Whether suspicious patterns were detected in user prompt |

Expand Down
123 changes: 64 additions & 59 deletions action.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -70,10 +70,6 @@ inputs:
description: "Enable yolo mode - auto-approve all prompts (true/false)"
required: false
default: "true"
quiet:
description: "Suppress verbose tool call output (true/false). Set to false for debugging."
required: false
default: "true"
max-retries:
description: "Maximum number of retries on failure (0 = no retries)"
required: false
Expand Down Expand Up @@ -107,6 +103,9 @@ outputs:
execution-time:
description: "Agent execution time in seconds"
value: ${{ steps.run-agent.outputs.execution-time }}
verbose-log-file:
description: "Path to the full verbose agent log (includes tool calls)"
value: ${{ steps.run-agent.outputs.verbose-log-file }}
security-blocked:
description: "Whether execution was blocked due to security concerns"
value: ${{ steps.sanitize-input.outputs.blocked == 'true' || steps.sanitize-output.outputs.leaked == 'true' }}
Expand Down Expand Up @@ -432,7 +431,6 @@ runs:
WORKING_DIR: ${{ inputs.working-directory }}
CAGENT_VERSION: ${{ inputs.cagent-version }}
MCP_INSTALLED: ${{ steps.setup-binaries.outputs.mcp-installed }}
QUIET: ${{ inputs.quiet }}
ADD_PROMPT_FILES: ${{ inputs.add-prompt-files }}
MAX_RETRIES: ${{ inputs.max-retries }}
RETRY_DELAY: ${{ inputs.retry-delay }}
Expand Down Expand Up @@ -461,10 +459,22 @@ runs:

# Create output file early (before any validation exits)
# This ensures downstream steps always have a valid output file reference
OUTPUT_FILE=$(mktemp /tmp/cagent-output.XXXXXX.log)
OUTPUT_FILE=$(mktemp /tmp/cagent-output-XXXXXX)
if [ -z "$OUTPUT_FILE" ] || [ ! -f "$OUTPUT_FILE" ]; then
echo "::error::Failed to create output file"
exit 1
fi
echo "output-file=$OUTPUT_FILE" >> $GITHUB_OUTPUT
echo "Output file: $OUTPUT_FILE"

VERBOSE_LOG_FILE=$(mktemp /tmp/cagent-verbose-XXXXXX)
if [ -z "$VERBOSE_LOG_FILE" ] || [ ! -f "$VERBOSE_LOG_FILE" ]; then
echo "::error::Failed to create verbose log file"
exit 1
fi
echo "verbose-log-file=$VERBOSE_LOG_FILE" >> $GITHUB_OUTPUT
echo "Verbose log file: $VERBOSE_LOG_FILE"

# Build command arguments array (SECURE: no eval!)
ARGS=("run" "--exec")

Expand All @@ -473,11 +483,6 @@ runs:
ARGS+=("--yolo")
fi

# Quiet mode: suppress verbose tool output (default: true)
if [ "$QUIET" = "true" ]; then
ARGS+=("--hide-tool-calls" "--hide-tool-results")
fi

# Set working directory so relative paths (e.g., memory toolset) resolve
# from the repo root, not from the agent YAML's parent directory
ARGS+=("--working-dir" "$(pwd)")
Expand Down Expand Up @@ -527,22 +532,25 @@ runs:
echo "🔄 Retry attempt $((ATTEMPT - 1)) of $MAX_RETRIES (waiting ${CURRENT_DELAY}s)..."
sleep "$CURRENT_DELAY"
CURRENT_DELAY=$((CURRENT_DELAY * 2))
# Reset output file for retry
# Reset clean output file; append a separator to verbose log to preserve earlier attempts
> "$OUTPUT_FILE"
echo "" >> "$VERBOSE_LOG_FILE"
echo "========== RETRY ATTEMPT $ATTEMPT ($(date -u +%Y-%m-%dT%H:%M:%SZ)) ==========" >> "$VERBOSE_LOG_FILE"
echo "" >> "$VERBOSE_LOG_FILE"
fi

# SECURE: Direct execution with quoted arguments (no eval!)
# Output goes to verbose log file only (keeps console clean)
# PIPESTATUS: [0]=printf [1]=cagent/timeout
set +e # Don't exit on command failure
# Pipeline: printf | [timeout] cagent | tee
# PIPESTATUS: [0]=printf [1]=cagent/timeout [2]=tee
if [ "$TIMEOUT" != "0" ]; then
printf '%s\n' "$PROMPT_INPUT" | timeout "$TIMEOUT" "$GITHUB_WORKSPACE/cagent" "${ARGS[@]}" 2>&1 | tee "$OUTPUT_FILE"
printf '%s\n' "$PROMPT_INPUT" | timeout "$TIMEOUT" "$GITHUB_WORKSPACE/cagent" "${ARGS[@]}" >> "$VERBOSE_LOG_FILE" 2>&1
EXIT_CODE=${PIPESTATUS[1]}
if [ $EXIT_CODE -eq 124 ]; then
echo "::error::Agent execution timed out after $TIMEOUT seconds"
fi
else
printf '%s\n' "$PROMPT_INPUT" | "$GITHUB_WORKSPACE/cagent" "${ARGS[@]}" 2>&1 | tee "$OUTPUT_FILE"
printf '%s\n' "$PROMPT_INPUT" | "$GITHUB_WORKSPACE/cagent" "${ARGS[@]}" >> "$VERBOSE_LOG_FILE" 2>&1
EXIT_CODE=${PIPESTATUS[1]}
fi
set -e
Expand All @@ -566,6 +574,28 @@ runs:
echo "::warning::Agent failed (exit code: $EXIT_CODE), will retry..."
done

# Produce clean output (strip tool calls/results) for downstream steps
awk '
/<thinking>/,/<\/thinking>/ { next }
/^\[thinking\]/,/^\[\/thinking\]/ { next }
/^Thinking:/ { next }
/^--- Tool:/ { in_tool=1; next }
in_tool && /^--- (Tool:|Agent:|$)/ { in_tool=0; next }
in_tool { next }
/^--- Agent:/ { next }
/^time=/ { next }
/^level=/ { next }
/^msg=/ { next }
/^> \[!NOTE\]/ { next }
/For any feedback/ { next }
/transfer_task/ { next }
/Delegating to/ { next }
/Task delegated/ { next }
NF==0 && !seen_content { next }
NF>0 { seen_content=1 }
{ print }
' "$VERBOSE_LOG_FILE" > "$OUTPUT_FILE"

END_TIME=$(date +%s)
EXECUTION_TIME=$((END_TIME - START_TIME))

Expand Down Expand Up @@ -629,58 +659,33 @@ runs:

$ACTION_PATH/security/sanitize-output.sh "$OUTPUT_FILE"

# Extract clean output (remove verbose cagent internals)
echo "🧹 Extracting clean agent output..."

# Primary method: Extract from cagent-output code block (most reliable)
# Extract from cagent-output code block if present (overrides awk-filtered output)
# Note: the code fence may not be at the start of a line if the agent
# emits conversational text before it, so we avoid anchoring with ^.
if grep -q '```cagent-output' "$OUTPUT_FILE"; then
echo "🧹 Extracting clean output from cagent-output code block..."
awk '
/```cagent-output/ { capturing=1; next }
capturing && /^```/ { capturing=0; next }
capturing && /```/ { capturing=0; next }
capturing { print }
' "$OUTPUT_FILE" > "${OUTPUT_FILE}.clean"
echo "✅ Extracted clean output from cagent-output code block"
else
# Use awk to filter out thinking blocks, tool calls, and internal markers
awk '
# Skip thinking blocks entirely
/<thinking>/,/<\/thinking>/ { next }
/^\[thinking\]/,/^\[\/thinking\]/ { next }
/^Thinking:/ { next }

# Skip tool call blocks
/^--- Tool:/ { in_tool=1; next }
in_tool && /^---/ { in_tool=0; next }
in_tool { next }

# Skip agent markers and metadata
/^--- Agent:/ { next }
/^time=/ { next }
/^level=/ { next }
/^msg=/ { next }
/^> \[!NOTE\]/ { next }
/For any feedback/ { next }

# Skip transfer_task and delegation output
/transfer_task/ { next }
/Delegating to/ { next }
/Task delegated/ { next }

# Skip empty lines at start
NF==0 && !seen_content { next }
NF>0 { seen_content=1 }

# Print everything else
{ print }
' "$OUTPUT_FILE" > "${OUTPUT_FILE}.clean"

echo "✅ Filtered verbose output"
if [ -s "${OUTPUT_FILE}.clean" ]; then
mv "${OUTPUT_FILE}.clean" "$OUTPUT_FILE"
echo "✅ Extracted clean output from cagent-output code block"
else
echo "::warning::Extracted cagent-output code block is empty, keeping filtered output"
rm -f "${OUTPUT_FILE}.clean"
fi
fi

# Use the cleaned output
mv "${OUTPUT_FILE}.clean" "$OUTPUT_FILE"
- name: Upload verbose agent log
if: always()
uses: actions/upload-artifact@ea165f8d65b6e75b540449e92b4886f43607fa02 # v4.6.2
with:
name: cagent-verbose-log-${{ github.run_id }}-${{ github.run_attempt }}
path: ${{ steps.run-agent.outputs.verbose-log-file }}
retention-days: 14
if-no-files-found: ignore

- name: Update job summary with cleaned output
if: always()
Expand Down
Loading