Discuss

/meridian:discuss — Phase Discussion

Structured context gathering and decision-making before planning. Identifies gray areas in phase scope and presents focused questions to capture implementation decisions.

Arguments

• `` — (optional) Phase ID to discuss. Defaults to current active phase.
• --auto — Skip interactive questions, apply recommended defaults.
• --chain — Run discuss then automatically advance to plan.
• --batch — Dump all questions to output for offline answering.
• --persona — Frame context-gathering through a role lens (pm, architect, ux, qa, security)

Keywords

discuss, context, gray areas, questions, decisions, scope, approach, before planning

Persona Support

If --persona is passed, load the persona and use it to frame the context-gathering questions. The persona shapes which gray areas are surfaced and how questions are phrased.

PYTHONPATH=$MERIDIAN_HOME uv run --project $MERIDIAN_HOME -- python -c "
import json
from scripts.personas import load_persona
persona = load_persona('')
print(json.dumps({'label': persona['label'], 'content': persona['content']}, indent=2))
"

Inject the persona content into the context-gatherer prompt before calling run_discuss. Show the active persona in the discuss banner:

## Discuss — 
Persona: 
Gray areas will be framed from the  perspective.

Available personas: pm, architect, ux, qa, security

Procedure

1. Load phase — Read phase from DB, parse acceptance criteria.

PYTHONPATH=$MERIDIAN_HOME uv run --project $MERIDIAN_HOME -- python -c "
import json, sys
from scripts.db import open_project
from scripts.state import get_status
with open_project('.') as conn:
    status = get_status(conn)
    print(json.dumps(status, default=str))
"

1. Run discuss — Identify gray areas, generate questions, optionally auto-apply.

PYTHONPATH=$MERIDIAN_HOME uv run --project $MERIDIAN_HOME -- python -c "
import json, sys
from pathlib import Path
from scripts.db import open_project
from scripts.discuss_phase import run_discuss
mode = sys.argv[1] if len(sys.argv) > 1 else 'interactive'
phase_id = int(sys.argv[2]) if len(sys.argv) > 2 else None
chain = '--chain' in sys.argv
with open_project('.') as conn:
    if not phase_id:
        from scripts.state import get_status
        status = get_status(conn)
        active = status.get('active_phase')
        phase_id = active['id'] if active else None
    if not phase_id:
        print('ERROR: No active phase found. Specify phase_id.')
        sys.exit(1)
    result = run_discuss(conn, phase_id, 'default', Path('.'), mode=mode, chain=chain)
    print(json.dumps(result, default=str, indent=2))
" -- "$MODE" "$PHASE_ID"

1. Interactive flow — If mode is interactive:

• Present each gray area with options
• Ask the user to choose (or accept defaults)
• After all questions answered, call apply_answers() to persist
• Write CONTEXT.md to .meridian/phases/

1. Apply answers — Persist decisions and generate context doc.

PYTHONPATH=$MERIDIAN_HOME uv run --project $MERIDIAN_HOME -- python -c "
import json, sys
from pathlib import Path
from scripts.db import open_project
from scripts.discuss_phase import apply_answers
phase_id = int(sys.argv[1])
answers_json = sys.argv[2]
gray_areas_json = sys.argv[3]
answers = json.loads(answers_json)
gray_areas = json.loads(gray_areas_json)
with open_project('.') as conn:
    result = apply_answers(conn, phase_id, 'default', gray_areas, answers, Path('.'))
    print(result['context_doc'])
"

1. Transition phase — Move phase to context_gathered state.

PYTHONPATH=$MERIDIAN_HOME uv run --project $MERIDIAN_HOME -- python -c "
from scripts.db import open_project
from scripts.state import transition_phase
phase_id = int('$PHASE_ID')
with open_project('.') as conn:
    transition_phase(conn, phase_id, 'context_gathered')
    print(f'Phase {phase_id} transitioned to context_gathered')
"

1. Chain advance — If --chain was passed, automatically invoke /meridian:plan for this phase.

Output

Display gray areas as numbered list with options. After decisions are made, show the generated CONTEXT.md summary. If --auto, show applied defaults.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: mattjaikaran
• Source: mattjaikaran/meridian
• License: Apache-2.0

Install and usage instructions live in the source repository linked above.