briefing

Name: briefing
Author: softspark

161stars21forks

How to install briefing?

softspark/ai-toolkit/briefing

$npx -y skills add softspark/ai-toolkit --skill briefing

Installs into the current project.

Summary

Executive daily briefing aggregating reports from all agents into decision-focused summary. Triggers: briefing, daily summary, status across system, executive update.

Files · 1

View on GitHub

SKILL.md

1	# Briefing Command
2
3	Triggers the Chief of Staff to generate an executive summary.
4
5	## Usage
6
7	```bash
8	/briefing [period]
9	# Example: /briefing today
10	# Example: /briefing week
11	/briefing --tokens [--since 7d\|24h\|30m]
12	# Reports real session token usage from Claude Code JSONL.
13	```
14
15	## Protocol
16	1. Collect: Gather logs from `kb/learnings/`, `maintenance/` logs, and recent runs.
17	2. Synthesize: Group by category (Ops, Strategy, Actions).
18	3. Filter: Remove low-priority success logs.
19	4. Present: Render the Daily Brief.
20
21	## Example
22
23	```
24	## Daily Brief — 2026-04-23
25
26	### Ops
27	- night-watch: 3 dep updates shipped, 1 rolled back (breaking change in `x-pkg@2.0`)
28	- health: all green except `mailpit` (degraded, non-critical)
29
30	### Strategy
31	- predict: new PR #42 overlaps with in-flight refactor in `/src/auth`
32
33	### Actions needed
34	- Review rollback from night-watch (ETA: 5 min)
35	- Decide on `x-pkg` pin strategy (open question on GitHub #41)
36	```
37
38	## Rules
39
40	- MUST stay under 200 words unless the user explicitly asks for more detail
41	- MUST lead with decision-relevant facts, not chronology — "what should I act on" before "what happened"
42	- NEVER invent agent activity — report only what the logs show; absence of logs means "no data", not "nothing happened"
43	- CRITICAL: separate Ops (what ran) from Strategy (what was decided) from Actions (what needs a human) — mixing them defeats the brief
44	- MANDATORY: when no material activity exists for a category, omit the category heading instead of writing "none"
45
46	## Gotchas
47
48	- `kb/learnings/` often mixes drafts with completed entries. Filter by frontmatter `status: final` or by filename convention before aggregating.
49	- `maintenance/` branch logs from `/night-watch` use a different format (Shift Report markdown) than agent run logs. Do not concatenate blindly — parse each source separately and normalize.
50	- "Recent runs" without an explicit time bound defaults to everything on some log backends. Always pass `--since` or a date filter, or you will read a week into yesterday's memory.
51	- Successful runs outnumber interesting runs by an order of magnitude. Aggressively filter green/noop entries — they are the signal's noise floor.
52
53	## Token Receipts
54
55	The `--tokens` flag reports real token usage parsed from Claude Code session JSONL — not estimates. Useful for:
56
57	- Verifying `output-mode: concise` actually reduces tokens vs default sessions
58	- Spotting expensive runs before they show up on the bill
59	- Capturing a baseline before changing prompts or skills
60
61	Underlying script: `scripts/session_token_stats.py`.
62
63	```bash
64	# Aggregate current session
65	python3 scripts/session_token_stats.py --json
66
67	# Statusline-friendly one-line output
68	python3 scripts/session_token_stats.py --statusline
69
70	# Trend vs baseline
71	python3 scripts/session_token_stats.py --statusline --baseline ~/.softspark/ai-toolkit/baseline.json
72	```
73
74	### Status line (installed by default in v3.2.0+)
75
76	`ai-toolkit install` wires `~/.claude/settings.json` to `app/hooks/ai-toolkit-statusline.sh`. The hook reads native Claude Code statusLine stdin (no session JSONL parsing) and renders one line:
77
78	```
79	➜ <dir> git:(branch) ✗ ████░░░░░░ 43% ↑6.5k ↓252k effort:xhigh <model>
80	```
81
82	Segments left to right:
83
84	- `➜ <dir>` — current directory basename
85	- `git:(branch) ✗` — git branch + dirty marker
86	- 10-cell progress bar for context-window usage. Color tiers: green `<70%`, orange `70–89%`, red `≥90%`
87	- `↑in ↓out` — token arrows. Green up = input (upload), red down = output (download). Both cumulative across the session.
88	- `effort:level` — Claude Code effort level (low / medium / high / xhigh)
89	- model name
90
91	Custom statusLine entries you set yourself (without the `_source: ai-toolkit` tag) are preserved untouched on install.
92
93	Opt-outs (no reinstall required):
94
95	- `AI_TOOLKIT_STATUSLINE_DISABLE=1` — silence the line entirely
96	- `AI_TOOLKIT_STATUSLINE_NO_TOKENS=1` — hide token arrows segment
97	- `AI_TOOLKIT_STATUSLINE_NO_GIT=1` — hide git segment
98	- `AI_TOOLKIT_STATUSLINE_NO_EFFORT=1` — hide effort level segment
99	- `AI_TOOLKIT_STATUSLINE_NO_COLOR=1` — disable ANSI colors
100	- `AI_TOOLKIT_STATUSLINE_SHOW_COST=1` — append Claude Code's reported cost (`cost.total_cost_usd`)
101	- `AI_TOOLKIT_STATUSLINE_DUMP=1` — write received stdin to `/tmp/cc-statusline-input.json` (debug)
102
103	### Save a baseline
104
105	```bash
106	python3 scripts/session_token_stats.py --json \| jq '.totals' > ~/.softspark/ai-toolkit/baseline.json
107	export AI_TOOLKIT_STATUSLINE_BASELINE=~/.softspark/ai-toolkit/baseline.json
108	```
109
110	The statusline then renders trend arrows (↑ / ↓) against that baseline.
111
112	## When NOT to Use
113
114	- For a specific production incident — use `/workflow incident-

People who liked this also viewed

6 picks

Type

briefing

161stars21forks

View on GitHub

How to install briefing?

softspark/ai-toolkit/briefing

$npx -y skills add softspark/ai-toolkit --skill briefing

Installs into the current project.

Summary

Executive daily briefing aggregating reports from all agents into decision-focused summary. Triggers: briefing, daily summary, status across system, executive update.

Files · 1

View on GitHub

SKILL.md

1	# Briefing Command
2
3	Triggers the Chief of Staff to generate an executive summary.
4
5	## Usage
6
7	```bash
8	/briefing [period]
9	# Example: /briefing today
10	# Example: /briefing week
11	/briefing --tokens [--since 7d\|24h\|30m]
12	# Reports real session token usage from Claude Code JSONL.
13	```
14
15	## Protocol
16	1. Collect: Gather logs from `kb/learnings/`, `maintenance/` logs, and recent runs.
17	2. Synthesize: Group by category (Ops, Strategy, Actions).
18	3. Filter: Remove low-priority success logs.
19	4. Present: Render the Daily Brief.
20
21	## Example
22
23	```
24	## Daily Brief — 2026-04-23
25
26	### Ops
27	- night-watch: 3 dep updates shipped, 1 rolled back (breaking change in `x-pkg@2.0`)
28	- health: all green except `mailpit` (degraded, non-critical)
29
30	### Strategy
31	- predict: new PR #42 overlaps with in-flight refactor in `/src/auth`
32
33	### Actions needed
34	- Review rollback from night-watch (ETA: 5 min)
35	- Decide on `x-pkg` pin strategy (open question on GitHub #41)
36	```
37
38	## Rules
39
40	- MUST stay under 200 words unless the user explicitly asks for more detail
41	- MUST lead with decision-relevant facts, not chronology — "what should I act on" before "what happened"
42	- NEVER invent agent activity — report only what the logs show; absence of logs means "no data", not "nothing happened"
43	- CRITICAL: separate Ops (what ran) from Strategy (what was decided) from Actions (what needs a human) — mixing them defeats the brief
44	- MANDATORY: when no material activity exists for a category, omit the category heading instead of writing "none"
45
46	## Gotchas
47
48	- `kb/learnings/` often mixes drafts with completed entries. Filter by frontmatter `status: final` or by filename convention before aggregating.
49	- `maintenance/` branch logs from `/night-watch` use a different format (Shift Report markdown) than agent run logs. Do not concatenate blindly — parse each source separately and normalize.
50	- "Recent runs" without an explicit time bound defaults to everything on some log backends. Always pass `--since` or a date filter, or you will read a week into yesterday's memory.
51	- Successful runs outnumber interesting runs by an order of magnitude. Aggressively filter green/noop entries — they are the signal's noise floor.
52
53	## Token Receipts
54
55	The `--tokens` flag reports real token usage parsed from Claude Code session JSONL — not estimates. Useful for:
56
57	- Verifying `output-mode: concise` actually reduces tokens vs default sessions
58	- Spotting expensive runs before they show up on the bill
59	- Capturing a baseline before changing prompts or skills
60
61	Underlying script: `scripts/session_token_stats.py`.
62
63	```bash
64	# Aggregate current session
65	python3 scripts/session_token_stats.py --json
66
67	# Statusline-friendly one-line output
68	python3 scripts/session_token_stats.py --statusline
69
70	# Trend vs baseline
71	python3 scripts/session_token_stats.py --statusline --baseline ~/.softspark/ai-toolkit/baseline.json
72	```
73
74	### Status line (installed by default in v3.2.0+)
75
76	`ai-toolkit install` wires `~/.claude/settings.json` to `app/hooks/ai-toolkit-statusline.sh`. The hook reads native Claude Code statusLine stdin (no session JSONL parsing) and renders one line:
77
78	```
79	➜ <dir> git:(branch) ✗ ████░░░░░░ 43% ↑6.5k ↓252k effort:xhigh <model>
80	```
81
82	Segments left to right:
83
84	- `➜ <dir>` — current directory basename
85	- `git:(branch) ✗` — git branch + dirty marker
86	- 10-cell progress bar for context-window usage. Color tiers: green `<70%`, orange `70–89%`, red `≥90%`
87	- `↑in ↓out` — token arrows. Green up = input (upload), red down = output (download). Both cumulative across the session.
88	- `effort:level` — Claude Code effort level (low / medium / high / xhigh)
89	- model name
90
91	Custom statusLine entries you set yourself (without the `_source: ai-toolkit` tag) are preserved untouched on install.
92
93	Opt-outs (no reinstall required):
94
95	- `AI_TOOLKIT_STATUSLINE_DISABLE=1` — silence the line entirely
96	- `AI_TOOLKIT_STATUSLINE_NO_TOKENS=1` — hide token arrows segment
97	- `AI_TOOLKIT_STATUSLINE_NO_GIT=1` — hide git segment
98	- `AI_TOOLKIT_STATUSLINE_NO_EFFORT=1` — hide effort level segment
99	- `AI_TOOLKIT_STATUSLINE_NO_COLOR=1` — disable ANSI colors
100	- `AI_TOOLKIT_STATUSLINE_SHOW_COST=1` — append Claude Code's reported cost (`cost.total_cost_usd`)
101	- `AI_TOOLKIT_STATUSLINE_DUMP=1` — write received stdin to `/tmp/cc-statusline-input.json` (debug)
102
103	### Save a baseline
104
105	```bash
106	python3 scripts/session_token_stats.py --json \| jq '.totals' > ~/.softspark/ai-toolkit/baseline.json
107	export AI_TOOLKIT_STATUSLINE_BASELINE=~/.softspark/ai-toolkit/baseline.json
108	```
109
110	The statusline then renders trend arrows (↑ / ↓) against that baseline.
111
112	## When NOT to Use
113
114	- For a specific production incident — use `/workflow incident-

People who liked this also viewed

6 picks

Type